Initial release

.gitattributes

@@ -33,3 +33,14 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+assets/showcase/english-agentic-01.mp3 filter=lfs diff=lfs merge=lfs -text
+assets/showcase/english-agentic-02.mp3 filter=lfs diff=lfs merge=lfs -text
+assets/showcase/english-agentic-03.mp3 filter=lfs diff=lfs merge=lfs -text
+assets/showcase/english-audiobook-01.mp3 filter=lfs diff=lfs merge=lfs -text
+assets/showcase/english-audiobook-02.mp3 filter=lfs diff=lfs merge=lfs -text
+assets/showcase/english-audiobook-03.mp3 filter=lfs diff=lfs merge=lfs -text
+assets/showcase/german-audiobook-01.mp3 filter=lfs diff=lfs merge=lfs -text
+cb0/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+cb1/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+cb2/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+cb3/tokenizer.json filter=lfs diff=lfs merge=lfs -text

COVERED_FILES.md

@@ -0,0 +1,49 @@
+# TontaubeV1 Covered Files Manifest
+
+This is the Covered Files Manifest referenced by the Tontaube Community Model
+License 1.0 supplied with this release. The paths below are relative to the
+repository root. Checksums use SHA-256.
+
+The listed files are designated as Model Materials under that license only to
+the extent that the relevant rights are owned by or licensable by the
+Licensor. The files incorporate Qwen3 material that remains subject to Apache
+License 2.0; see [`THIRD_PARTY_NOTICES.md`](THIRD_PARTY_NOTICES.md).
+
+| Path | SHA-256 |
+|---|---|
+| `cb0/added_tokens.json` | `37708d972d9329be8bf6e78697622bb0f364fdb079eadc5f8ecde2afb0e594f7` |
+| `cb0/config.json` | `c11ec92b0205d2c0e4b8aad32c2832ee816f0dc21bd1b8b2fd7e1d8ed98daa40` |
+| `cb0/merges.txt` | `8831e4f1a044471340f7c0a83d7bd71306a5b867e95fd870f74d0c5308a904d5` |
+| `cb0/model.safetensors` | `17f990196944ada19734e43c195eaf998c998173edba00d14f01a4668865e814` |
+| `cb0/special_tokens_map.json` | `86cdf52f6d9fcc5c35c169b625730db8e178e1316c3c13258c2f665bf2ea1c2e` |
+| `cb0/tokenizer.json` | `8c838573f03edfd45dce6890c0ed9cb9ea51a3721f7385f6072c8bd2919d37ed` |
+| `cb0/tokenizer_config.json` | `c22893fb21d0430b099c67c9a85904f657f3dde3c0c0f045a0348ecf33fe5976` |
+| `cb0/vocab.json` | `ca10d7e9fb3ed18575dd1e277a2579c16d108e32f27439684afa0e10b1440910` |
+| `cb1/added_tokens.json` | `37708d972d9329be8bf6e78697622bb0f364fdb079eadc5f8ecde2afb0e594f7` |
+| `cb1/config.json` | `a4c535b45026418ca17f224b6d185d5f7596e8c3403b85777a0c639598364092` |
+| `cb1/merges.txt` | `8831e4f1a044471340f7c0a83d7bd71306a5b867e95fd870f74d0c5308a904d5` |
+| `cb1/model.safetensors` | `a2404b2b1f0ab6e92e82047e06d273821f3b413fedb8ad27b234d5526bf281c1` |
+| `cb1/special_tokens_map.json` | `86cdf52f6d9fcc5c35c169b625730db8e178e1316c3c13258c2f665bf2ea1c2e` |
+| `cb1/tokenizer.json` | `8c838573f03edfd45dce6890c0ed9cb9ea51a3721f7385f6072c8bd2919d37ed` |
+| `cb1/tokenizer_config.json` | `c22893fb21d0430b099c67c9a85904f657f3dde3c0c0f045a0348ecf33fe5976` |
+| `cb1/vocab.json` | `ca10d7e9fb3ed18575dd1e277a2579c16d108e32f27439684afa0e10b1440910` |
+| `cb2/added_tokens.json` | `37708d972d9329be8bf6e78697622bb0f364fdb079eadc5f8ecde2afb0e594f7` |
+| `cb2/config.json` | `9a73ee3e034c97eef4d29df0190445302fb54c514757a906588f1394cc0d2240` |
+| `cb2/merges.txt` | `8831e4f1a044471340f7c0a83d7bd71306a5b867e95fd870f74d0c5308a904d5` |
+| `cb2/model.safetensors` | `78fd0709da355a64438505e726ecaa49ecfb41c4a096298d8b88fd7566baa081` |
+| `cb2/special_tokens_map.json` | `86cdf52f6d9fcc5c35c169b625730db8e178e1316c3c13258c2f665bf2ea1c2e` |
+| `cb2/tokenizer.json` | `8c838573f03edfd45dce6890c0ed9cb9ea51a3721f7385f6072c8bd2919d37ed` |
+| `cb2/tokenizer_config.json` | `c22893fb21d0430b099c67c9a85904f657f3dde3c0c0f045a0348ecf33fe5976` |
+| `cb2/vocab.json` | `ca10d7e9fb3ed18575dd1e277a2579c16d108e32f27439684afa0e10b1440910` |
+| `cb3/added_tokens.json` | `37708d972d9329be8bf6e78697622bb0f364fdb079eadc5f8ecde2afb0e594f7` |
+| `cb3/config.json` | `07b897947d78e08a5a2076b2a8db9668697f8214c9b2190383ac33de4b0c508c` |
+| `cb3/merges.txt` | `8831e4f1a044471340f7c0a83d7bd71306a5b867e95fd870f74d0c5308a904d5` |
+| `cb3/model.safetensors` | `09dab1fb894af2cd98ff8a2bf784700a78ca67be749144fb02a7feb947d779bf` |
+| `cb3/special_tokens_map.json` | `86cdf52f6d9fcc5c35c169b625730db8e178e1316c3c13258c2f665bf2ea1c2e` |
+| `cb3/tokenizer.json` | `8c838573f03edfd45dce6890c0ed9cb9ea51a3721f7385f6072c8bd2919d37ed` |
+| `cb3/tokenizer_config.json` | `c22893fb21d0430b099c67c9a85904f657f3dde3c0c0f045a0348ecf33fe5976` |
+| `cb3/vocab.json` | `ca10d7e9fb3ed18575dd1e277a2579c16d108e32f27439684afa0e10b1440910` |
+
+`README.md`, `LICENSE`, `THIRD_PARTY_NOTICES.md`, the files under
+`THIRD_PARTY_LICENSES/`, and this manifest are accompanying documentation, not
+Model Materials designated by this manifest.

LICENSE

@@ -0,0 +1,493 @@
+# Tontaube Community Model License 1.0
+
+Effective date: 26 August 2026
+
+## In short
+
+*This summary is not part of the Agreement and does not change it.*
+
+- **Research and noncommercial use is free**, without a revenue limit, subject
+  to the service and use restrictions below.
+- **Commercial use is free** while your group's revenue is **below EUR 200,000**.
+  At or above that amount, a separate commercial license is required after any
+  transition period under Section 4 — see <https://tontaube.ai/licensing/>.
+- **Testing is always free**, at any size, as long as it is not production.
+- **Offering the model to others as a speech-generation service** always needs a
+  commercial license, whatever your revenue. A small free demo is fine. Using it
+  inside your own product is different — the revenue rule above applies as normal.
+- **You may build on it, fine-tune it, and distill it.** If you distribute the
+  resulting weights, they carry this same license.
+- **Community commercial use requires attribution:** put "Powered by TontaubeV1"
+  somewhere reasonable — a credits line or a docs page is enough.
+
+---
+
+This Tontaube Community Model License 1.0 (the **Agreement**) is offered by
+Cremer & Cremer Technologies UG (haftungsbeschränkt), with registered office
+in Berlin, Germany, registered with the Commercial Register of Amtsgericht
+Charlottenburg under HRB 272354 B (**Licensor**), to the individual or legal
+entity exercising rights under this Agreement (**You** or **Licensee**).
+Tontaube is a brand of the Licensor and a registered European Union trade mark
+(EUTM no. 019205218). The Licensor's current legal notice, business address, and
+contact details are available at <https://tontaube.ai/impressum/>. You accept
+this Agreement by affirmatively indicating acceptance or by exercising any right
+granted only under this Agreement. Mere access to or receipt of Model Materials,
+without exercising a licensed right, does not by itself constitute acceptance. If
+You act for an organization, You represent that You have authority to bind it.
+
+## 1. Scope and definitions
+
+**Covered Files Manifest** means the manifest supplied with a release that
+identifies by file name or path, and where practicable by checksum, the files
+that Licensor designates as subject to this Agreement. If a release is published
+without a manifest, or if a copy You receive does not include one, Model
+Materials means all model weight and checkpoint files published by Licensor in
+that release.
+
+**Model Materials** means only the TontaubeV1 trained model weights,
+checkpoints, model-specific configuration files, and other model artifacts
+identified in the Covered Files Manifest, including the four TontaubeV1
+codebook checkpoints. This Agreement grants rights only to the extent those
+rights are owned by or licensable by Licensor.
+
+Model Materials do **not** include inference source code, the TontaubeV1
+verbalization checkpoint (separately licensed under the Apache License 2.0), or
+any third-party software, models, weights, tokenizers, codecs, documentation, or
+other components, including Qwen3, DualCodec, VibeVoice, or w2v-BERT 2.0
+components. Those materials remain subject to their respective licenses. If an
+artifact contains both Licensor-owned and third-party material, this Agreement
+applies only to Licensor's rights, and all applicable third-party terms continue
+to apply.
+
+**Derivative Model** means any model, model weights, adapter, checkpoint, or
+other model artifact derived directly or indirectly from, incorporating, or
+modifying or adapting the Model Materials or another Derivative Model,
+including a fine-tune, LoRA or other adapter, merge, pruned or quantized
+version, or converted checkpoint. Outputs are not Derivative Models. For the
+purposes of the conditions in this Agreement only, a model that generates speech
+and was trained, fine-tuned, distilled, or otherwise improved using Outputs is
+treated as a Derivative Model, whether or not it constitutes a derivative work
+under applicable law. This classification does not give Licensor ownership of
+that model or grant rights in contributions not owned by or licensable by
+Licensor.
+
+**Output** means audio or other content generated through use of the Model
+Materials or a Derivative Model.
+
+**Affiliate** means an entity that directly or indirectly controls, is
+controlled by, or is under common control with Licensee. **Control** means
+direct or indirect ownership of more than 50 percent of the voting interests
+or the legal power to direct an entity's management and policies. Entities
+controlled by the same natural person or persons are under common control with
+one another.
+
+**Consolidated Gross Revenue** means the gross revenue of Licensee and all
+Affiliates, from all sources, for the most recently completed fiscal year, or
+for the trailing twelve months if that amount is greater, calculated on a
+consolidated basis under the accounting principles used in their ordinary
+financial records. Intercompany amounts are counted only once. VAT, sales and
+similar transaction taxes collected for a government, and customer refunds, are
+excluded. Where Licensee is a natural person, only business and self-employment
+turnover is counted; income from employment is excluded. Where Licensee does not
+measure revenue, including a public body, foundation, or nonprofit organization,
+total annual budget or expenditure applies instead. Revenue denominated in
+another currency must be converted to euros using the European Central Bank
+reference rate, or the nearest reasonably available equivalent, on the last day
+of the relevant period.
+
+**Commercial Use** means use primarily intended for commercial advantage,
+business benefit, or monetary compensation. It includes offering a hosted
+service or API, client or contract work, use in a product or service,
+distribution in connection with a business, and use in internal business
+operations, whether or not the Model Materials are separately priced. Evaluation
+under Section 2a is not Commercial Use.
+
+**Research and Noncommercial Use** means personal, educational, evaluation,
+academic, or scientific use that is not primarily intended for commercial
+advantage, business benefit, or monetary compensation. Research intended for
+deployment in a commercial product, service, client engagement, or internal
+business operation is Commercial Use, subject to Section 2a.
+
+**Speech Generation Service** means providing third parties with hosted or
+remote access to speech-generation functionality powered by the Model Materials
+or a Derivative Model, where generating speech is itself the product, service, or
+a separately offered functionality, whether or not for a fee. Using the Model
+Materials or a Derivative Model locally, or on infrastructure operated solely for
+You or on Your behalf without providing third parties access to that
+functionality, is not a Speech Generation Service. Nor is using them solely as an
+internal component of a product or service whose primary functionality is not
+speech generation and that does not provide speech generation as a standalone,
+API-accessible, or separately offered feature.
+
+## 2. Research and noncommercial license
+
+Subject to this Agreement, Licensor grants You a worldwide, non-exclusive,
+non-transferable, non-sublicensable, royalty-free license, under
+intellectual-property rights owned or licensable by Licensor, to use, reproduce,
+host, modify, fine-tune, quantize, convert, create Derivative Models from, and
+redistribute the Model Materials and Derivative Models for Research and
+Noncommercial Use.
+
+This Section 2 is not conditioned on Licensee's revenue.
+
+## 2a. Evaluation
+
+Subject to this Agreement, Licensor grants You the rights stated in Section 2,
+other than the right to redistribute the Model Materials or Derivative Models,
+solely for testing and evaluating them, regardless of Consolidated Gross Revenue
+and regardless of whether You intend eventual Commercial Use. This grant is not
+limited in time and requires no registration.
+
+This grant does not extend to production use. Production use means serving real
+users, producing Output used in a product or service, or supporting live
+business operations, whether internal or external.
+
+## 3. Community commercial license
+
+Subject to this Agreement, Licensor grants You, for Commercial Use, the rights
+stated in Section 2, for so long as Consolidated Gross Revenue is less than
+EUR 200,000.
+
+The threshold applies to Licensee and its Affiliates in aggregate and to their
+revenue from all sources, whether or not that revenue was generated using the
+Model Materials. Investment proceeds are not revenue unless they are treated
+as revenue in Licensee's ordinary financial records.
+
+The license granted to You does not independently license an Affiliate. You
+may permit Your personnel, contractors, and cloud or other service providers
+to exercise Your rights solely on Your behalf, for Your benefit, and within
+the scope of Your license, provided they receive no independent rights in the
+Model Materials, are bound by written terms that protect the Model Materials,
+and comply with this Agreement. You are responsible for their acts and
+omissions relating to the Model Materials as if they were Your own. An
+Affiliate must independently accept and comply with this Agreement unless it
+acts solely as such a service provider on Your behalf.
+
+Where You perform work for a client and provide only completed Outputs, Your
+Consolidated Gross Revenue alone determines Your eligibility under this Section.
+The client need not accept this Agreement solely to receive or use those
+Outputs, except for model-training use under Section 7. Providing completed
+Outputs under this paragraph is not, by itself, a Speech Generation Service.
+
+This paragraph does not authorize providing the client with Model Materials or
+a Derivative Model, operating a Speech Generation Service, or entering into an
+arrangement structured principally to give the client the functional equivalent
+of such access or to avoid this Agreement's commercial-license requirements. A
+client that receives Model Materials or a Derivative Model is a recipient under
+Section 5 and must independently accept and comply with this Agreement.
+
+## 3a. Speech Generation Services and public demonstrations
+
+No license granted under this Agreement permits You to operate a Speech
+Generation Service, other than a Public Demonstration. Operating a Speech
+Generation Service requires a separate written commercial license, regardless of
+Consolidated Gross Revenue and whether the service is offered for a fee. For a
+separate commercial license, see <https://tontaube.ai/licensing/>.
+
+A **Public Demonstration** means a free, non-production interface made available
+solely to allow users to evaluate the Model Materials or a Derivative Model. It
+must provide no API or automated access, no service-level commitment, no
+integration into another product, service, or workflow, and no bulk use, and must
+impose reasonable per-user rate and output limits.
+
+## 4. Crossing the commercial threshold
+
+If Consolidated Gross Revenue reaches or exceeds EUR 200,000, Section 3 does not
+authorize You to make Commercial Use of the Model Materials or a Derivative
+Model, except as provided in this Section. If it first reaches or exceeds that
+amount while You are making authorized Commercial Use, Your permission for that
+Commercial Use continues until 30 days after the end of the calendar month in
+which the threshold was first reached or exceeded. Any Commercial Use after
+that requires a separate written commercial license that has become effective.
+For a separate commercial license, see <https://tontaube.ai/licensing/> or
+contact <mailto:contact@craitech.io>.
+
+If Consolidated Gross Revenue later falls below EUR 200,000, Section 3
+authorizes Your Commercial Use again, without further action by either party.
+
+Your rights for Research and Noncommercial Use under Section 2 and for
+evaluation under Section 2a continue, provided You comply with this Agreement.
+Crossing the threshold is not itself a breach, does not require deletion of the
+Model Materials or Derivative Models, and does not affect compliant rights
+previously granted to downstream recipients.
+
+Products, copies, or units that You distributed while authorized remain licensed
+and need not be recalled, withdrawn, or disabled. Further distribution in the
+course of Commercial Use after Your permission under Section 3 ends requires a
+separate written commercial license.
+
+Licensor and Licensee are not bound by a separate commercial license unless
+they enter into one in writing. Nothing in this Agreement obliges either party
+to agree to separate commercial terms.
+
+Licensor may, on reasonable written request and no more than once in any
+12-month period, ask You to confirm in writing whether Your Consolidated Gross
+Revenue has reached or exceeded the threshold. This is not an audit right and
+requires no disclosure of figures.
+
+## 5. Redistribution and notices
+
+You may redistribute the Model Materials or a Derivative Model only if You:
+
+1. include a copy of this Agreement, or a clear and prominent reference to it
+   identifying it by name and giving its URL;
+2. include the Covered Files Manifest You received, where one was supplied;
+3. preserve all copyright, license, and attribution notices supplied with the
+   Model Materials, including the required notice below;
+4. clearly identify any modifications You made; and
+5. do not state or imply that Licensor endorses Your modification, product, or
+   service.
+
+Required notice:
+
+> TontaubeV1 Model Materials. Copyright © 2026 Cremer & Cremer Technologies
+> UG (haftungsbeschränkt). Licensed under the Tontaube Community Model License
+> 1.0.
+
+A person who receives Model Materials without a Covered Files Manifest still
+holds a license under this Agreement, and the fallback in Section 1 determines
+its scope.
+
+**Products containing the Model Materials.** Where the Model Materials or a
+Derivative Model are embedded in a product supplied to end users, such as an
+application, a game, or a device, an end user who acquires that product is not a
+recipient for the purposes of this Section, and no notice is owed to them under
+it. Attribution under Section 5a is Your obligation as the supplier, not theirs.
+Where You supplied those units or copies while
+licensed under this Agreement, they are licensed for their purchaser's use of the
+model as embodied in the product, permanently, and independently of whether Your
+own license later ends. Their purchaser requires no license under this Agreement
+for that use, and a purchaser who acquired the product in the ordinary course of
+business may rely on this unless it knew that You were not licensed.
+
+This paragraph does not apply where generating speech is the principal function of
+the product, or where an arrangement is structured principally to avoid the
+requirement for a commercial license. In those cases the purchaser requires its
+own license.
+
+Each recipient receives rights directly from Licensor under this Agreement
+only with respect to rights owned by or licensable by Licensor in the Model
+Materials and must independently satisfy its conditions, including the
+commercial revenue threshold. You must separately grant any rights needed for
+Your own contributions to a Derivative Model; Licensor does not license those
+contributions on Your behalf. You may impose terms on Your own contributions,
+products, or services, but may not represent that those terms replace or
+reduce a recipient's obligations under this Agreement for the Model Materials.
+
+Providing only Outputs does not distribute Model Materials or a
+Derivative Model. Except where a recipient affirmatively accepts this Agreement
+for model-training use under Section 7, a person who receives only Outputs need
+not accept this Agreement.
+
+## 5a. Attribution
+
+Where You make Commercial Use of the Model Materials or a Derivative Model under
+Section 3, You must state that fact using the wording "Powered by TontaubeV1".
+The statement may appear in product documentation, an about or credits screen, a
+service website, or any comparable place where a user or reader would reasonably
+find it. No particular placement, prominence, or prior approval is required, and
+where a product offers no such place, no statement is required.
+
+This Section does not apply to Commercial Use authorized by a separate written
+commercial license, unless that license provides otherwise. It does not apply to
+a person who acquires and uses a product containing the Model Materials without
+themselves redistributing them; attribution for such a product is the obligation
+of the party that supplied it.
+
+Where You distribute a Derivative Model, You must state in its documentation or
+model card that it was derived from or trained using TontaubeV1.
+
+Licensor grants You a limited, non-exclusive, royalty-free license to use the "Tontaube" and
+"TontaubeV1" word marks solely to make the statements required by this Section,
+in unmodified form and without implying endorsement. This license lasts for as
+long as those statements are required of You and You comply with them. You must not
+use "Tontaube" in the name of a Derivative Model, product, or service, or as
+branding of Your own, without Licensor's prior written permission. Nothing in
+this Section restricts use of the marks that applicable law permits without a
+license, such as referring to the Model Materials to describe their origin or
+Your product's compatibility.
+
+## 6. Outputs
+
+Licensor claims no ownership in an Output solely because it was generated
+using the Model Materials, to the extent permitted by applicable law. This
+Agreement does not determine whether an Output is protected by intellectual-
+property rights, who may own such rights, or whether an Output is lawful or
+non-infringing, and it grants no rights in third-party material contained in an
+Output. Licensee is responsible for Outputs and their use, including obtaining
+permissions and complying with rights of speakers and other third parties.
+
+Outputs generated in compliance with this Agreement while Licensee was
+authorized remain usable, subject to applicable law and third-party rights,
+after Licensee reaches or exceeds the commercial threshold or this Agreement
+terminates.
+No provision of this Agreement requires deletion of those Outputs. The
+obligations in Section 7 governing the use or supply of Outputs for model
+training continue to apply.
+
+## 7. Use conditions
+
+You must use the Model Materials, Derivative Models, and Outputs in compliance
+with applicable law, including export control and sanctions law, and must not
+knowingly infringe or facilitate infringement of another person's privacy,
+publicity, personality, copyright, trademark, or other rights.
+
+You must not knowingly use the Model Materials or a Derivative Model, or
+knowingly provide a service designed to enable others to use them, to:
+
+1. commit or materially facilitate fraud; or
+2. imitate an identifiable natural person's voice without that person's
+   authorization in a manner reasonably likely to cause a listener to believe
+   that synthetic speech is an authentic statement by that person.
+
+The second restriction does not prohibit authorized performances or lawful
+parody, satire, commentary, journalism, or public-interest research that is
+not reasonably likely to deceive a listener about the speech's authenticity.
+
+**Models trained on the Model Materials or Outputs.** Subject to the scope and
+eligibility requirements of the applicable license under Section 2, 2a, or 3, You
+may use the Model Materials, a Derivative Model, or Outputs to train, fine-tune,
+distill, or otherwise improve another model, and You may publish papers,
+evaluation results, samples, and other findings arising from that work. If You distribute or
+otherwise make available the weights, checkpoints, or adapters of a model so
+trained, and that model generates speech, You must distribute it under this
+Agreement and state in its documentation or model card that it was trained using
+TontaubeV1 Model Materials or Outputs.
+
+You must not knowingly provide or make Outputs available as training data,
+including in bulk or as a dataset, or otherwise for the stated or reasonably
+apparent purpose of training, fine-tuning, distilling, or improving a
+speech-generation, voice-conversion, or dubbing model, unless, before receiving
+them, the recipient affirmatively accepts this Agreement for that training use or
+enters into a separate written agreement with Licensor authorizing that use. This
+obligation applies to You as the supplier of the Outputs. It does not restrict
+ordinary publication, distribution, receipt, or listening where the Outputs are
+not supplied for model-training purposes.
+
+This requirement does not apply to a model whose function is speech enhancement,
+denoising, or restoration, or to any purpose that does not produce generated
+speech, including automatic speech recognition, diarization, speaker
+recognition, audio classification, benchmarking, and evaluation. If You
+distribute or otherwise make available the weights, checkpoints, or adapters of
+such a model, the statement of origin described above is required, but this
+Agreement does not require You to license that model under this Agreement. Voice
+conversion and dubbing models are subject to the requirement.
+
+These use conditions are part of this fixed version of the Agreement. No
+separately published or later-modified acceptable-use policy is incorporated
+by reference.
+
+## 8. Ownership, patents, and trademarks
+
+Licensor and its licensors retain their respective rights in the Model
+Materials. You retain whatever rights You may have in Your original
+contributions to a modification or Derivative Model, subject to the underlying
+Model Materials and applicable third-party rights. This Agreement does not
+determine ownership of a modification or Derivative Model as a whole.
+
+Subject to this Agreement, Licensor grants You a worldwide, non-exclusive,
+royalty-free, irrevocable patent license, under patent claims owned or
+licensable by Licensor that are necessarily infringed by the Model Materials as
+distributed by Licensor, to make, use, offer to sell, sell, import, and
+otherwise transfer the Model Materials. This license does not extend to Your
+modifications or to any combination of the Model Materials with other material.
+If You initiate patent litigation against any entity alleging that the Model
+Materials or a Derivative Model constitutes patent infringement, the patent
+license granted to You under this Section terminates on the date the litigation
+is filed.
+
+Except as stated in Section 5a, no trademark, trade-name, logo, or branding
+rights are granted.
+
+## 9. No warranty
+
+Under this Agreement, the Model Materials are provided free of charge and
+**as is**, without any guarantee of availability, accuracy, fitness for a
+particular purpose, or non-infringement, to the maximum extent permitted by
+applicable law. A separate written agreement signed by Licensor may expressly
+provide different warranties or support. You are solely responsible for
+evaluating whether the Model Materials, Derivative Models, and Outputs are
+appropriate and lawful for Your use.
+
+The Model Materials are not designed, tested, or validated for safety-critical
+use, including medical, clinical, emergency, aviation, automotive safety, or
+industrial control applications. Generated speech may omit, repeat, mispronounce,
+or alter text. Any such use is at Licensee's own risk and responsibility.
+
+## 10. Liability
+
+Licensor's liability is unlimited for intent and gross negligence; injury to
+life, body, or health; liability under mandatory product-liability law; and any
+express guarantee or other liability that cannot lawfully be limited.
+
+For slight negligence, Licensor is liable only for breach of a material
+contractual obligation whose performance is essential to this Agreement and
+on which Licensee may reasonably rely. In that case, liability is limited to
+the foreseeable damage typical for this Agreement. In all other cases,
+Licensor's liability for slight negligence is excluded to the extent permitted
+by law.
+
+Where Licensee is not a consumer and makes Commercial Use of the Model
+Materials, Licensee shall indemnify Licensor against third-party claims arising
+from Licensee's breach of Section 7, except to the extent the claim results from
+Licensor's own intent or gross negligence.
+
+## 11. Term and termination
+
+This Agreement begins when You first accept it or exercise rights under it and
+continues until terminated. You may terminate it at any time by permanently
+ceasing to exercise the rights it grants. Licensor may terminate this
+Agreement only for Your breach. For a remediable breach, Licensor must give
+written notice describing the breach and at least 30 calendar days to cure it
+before termination. A breach is non-remediable only where cure is impossible
+as a matter of fact or law; such a breach may be terminated by written notice
+with immediate effect.
+
+Licensor may terminate this Agreement by written notice with immediate effect
+for breach of Section 7, paragraph 2 (fraud and unauthorized voice imitation).
+All other breaches, including of the remaining use conditions, are subject to
+the cure period above.
+
+Upon termination, You must cease exercising rights in the Model Materials and
+Derivative Models that were granted by this Agreement, except for the patent
+license in Section 8. Termination does not require deletion of Outputs generated
+in compliance with this Agreement before termination, does not terminate a
+compliant downstream recipient's independent license, and does not affect
+products, copies, or units distributed while You were authorized.
+
+Sections 1, 5, 5a, 6, 8, 9, 10, 11, 12, and 13 survive termination to the extent
+necessary to give them effect, as does Section 7 solely as it applies to Outputs
+generated before termination.
+
+## 12. Static and later license versions
+
+This is a static, versioned Agreement. Licensor may offer the Model Materials
+under additional or different terms in the future, but doing so does not
+withdraw or modify rights already granted under this version.
+
+A later license version applies to You only if it expressly identifies itself
+as an optional replacement for this Agreement and You affirmatively elect to
+use the Model Materials under that later version. This Agreement does not
+grant rights under "any later version," and later versions do not apply
+automatically.
+
+## 13. General terms
+
+This Agreement is the entire agreement between Licensor and Licensee regarding
+the license granted here, except for a separate written agreement signed by
+both parties. A waiver must be in writing and applies only to the stated
+instance. If a provision is invalid or unenforceable, the remaining provisions
+remain in effect. To the extent required by applicable law, statutory
+provisions apply in place of the invalid or unenforceable provision; this
+Agreement does not itself create a replacement provision.
+
+This Agreement is governed by the laws of the Federal Republic of Germany,
+excluding its conflict-of-law rules and the United Nations Convention on
+Contracts for the International Sale of Goods. If Licensee is a merchant,
+public-law legal entity, or public-law special fund, the exclusive venue is
+Berlin, Germany. Otherwise, mandatory statutory venue rules apply.
+
+Nothing in this Agreement deprives a consumer of protections that cannot
+lawfully be waived by contract, including any mandatory protections of the
+country in which the consumer has their habitual residence. The choice of law
+and venue above is subject to those protections.

README.md

@@ -0,0 +1,282 @@
+---
+license: other
+license_name: tontaube-community-model-license-1.0
+license_link: https://huggingface.co/TontaubeAI/TontaubeV1/blob/main/LICENSE
+pipeline_tag: text-to-speech
+base_model:
+  - Qwen/Qwen3-1.7B
+  - Qwen/Qwen3-0.6B
+language:
+  - en
+  - de
+  - es
+  - fr
+  - it
+  - nl
+  - pt
+tags:
+  - text-to-speech
+  - voice-cloning
+  - multilingual
+  - audio
+  - vllm
+inference: false
+---
+
+<p align="center">
+  <img src="assets/tontaube-logo.png" alt="Tontaube" width="180">
+</p>
+
+# TontaubeV1
+
+TontaubeV1 is a multilingual text-to-speech model designed for expressive
+voice cloning, long-form generation, and low-latency streaming. English and
+German are its primary languages; Spanish, French, Italian, Dutch, and
+Portuguese are also supported.
+
+The release contains four causal codebook predictors. CB0 generates semantic
+audio and duration from text, while progressively smaller CB1--CB3 models add
+acoustic detail. Spoken text uses character-level tokenization, with compact
+BPE-encoded language and speaking-style controls. The public styles are
+`audiobook`, `conversational`, and `agentic`.
+
+[Technical report](https://tontaube.ai/papers/tontaube-v1-technical-report.pdf) ·
+[Try TontaubeV1 online](https://tontaube.ai/playground) ·
+[Inference repository](https://github.com/craitech/tontaube)
+
+## Quick start
+
+These checkpoints require the Tontaube vLLM model adapter and input layout;
+they are not drop-in `AutoModelForCausalLM` checkpoints. Use the matching
+release of the inference repository:
+
+```bash
+git clone --branch v1.0.0 --depth 1 \
+  https://github.com/craitech/tontaube.git
+cd tontaube
+uv run --python 3.12 --frozen tontaube serve
+```
+
+Once the API is ready, generate a WAV with the bundled English voice from a
+second terminal:
+
+```bash
+curl --silent --show-error --fail http://127.0.0.1:8080/predict \
+  -H 'content-type: application/json' \
+  -d '{"text":"Welcome to Tontaube.","language":"english","tag":"conversational","format":"wav"}' \
+| python3 -c 'import base64,json,sys; sys.stdout.buffer.write(base64.b64decode(json.load(sys.stdin)["audio_b64"]))' \
+> welcome.wav
+```
+
+Clone a voice by sending about 5--60 seconds of clean, single-speaker speech
+with the request:
+
+```bash
+VOICE_FILE=/path/to/reference.wav
+python3 -c 'import base64,json,sys; print(json.dumps({"text":"This voice was cloned from a local reference.","language":"english","tag":"conversational","voice_audio_b64":base64.b64encode(open(sys.argv[1],"rb").read()).decode(),"format":"wav"}))' "$VOICE_FILE" \
+| curl --silent --show-error --fail http://127.0.0.1:8080/predict \
+  -H 'content-type: application/json' \
+  --data-binary @- \
+| python3 -c 'import base64,json,sys; sys.stdout.buffer.write(base64.b64decode(json.load(sys.stdin)["audio_b64"]))' \
+> cloned.wav
+```
+
+The server downloads this model and its pinned runtime dependencies into local
+caches on first use. Follow the inference repository's installation
+instructions for system requirements, the browser interface, and Docker. The
+optional English verbalizer is available separately from
+`TontaubeAI/TontaubeV1-Verbalizer` and is downloaded only when enabled.
+
+The default semantic sampling temperature is `0.8`. A value around `0.6`
+generally favors text correctness, while values approaching `1.0` can produce
+more varied and expressive delivery.
+
+## Capabilities
+
+- Bundled synthetic voices and zero-shot cloning from up to 60 seconds of
+  reference audio
+- `audiobook`, `conversational`, and `agentic` speaking styles
+- Bounded-context long-form generation
+- WAV generation and low-latency MP3 or Opus streaming
+- Optional English text verbalizer, distributed separately
+
+## Examples
+
+These MP3 samples were normalized to a consistent speech listening level.
+
+### Audiobook
+
+**English 1**
+
+<audio controls preload="none" src="assets/showcase/english-audiobook-01.mp3"></audio>
+
+[Download MP3](assets/showcase/english-audiobook-01.mp3)
+
+**English 2**
+
+<audio controls preload="none" src="assets/showcase/english-audiobook-02.mp3"></audio>
+
+[Download MP3](assets/showcase/english-audiobook-02.mp3)
+
+**English 3**
+
+<audio controls preload="none" src="assets/showcase/english-audiobook-03.mp3"></audio>
+
+[Download MP3](assets/showcase/english-audiobook-03.mp3)
+
+**German 1**
+
+<audio controls preload="none" src="assets/showcase/german-audiobook-01.mp3"></audio>
+
+[Download MP3](assets/showcase/german-audiobook-01.mp3)
+
+### Agentic
+
+**English 1**
+
+<audio controls preload="none" src="assets/showcase/english-agentic-01.mp3"></audio>
+
+[Download MP3](assets/showcase/english-agentic-01.mp3)
+
+**English 2**
+
+<audio controls preload="none" src="assets/showcase/english-agentic-02.mp3"></audio>
+
+[Download MP3](assets/showcase/english-agentic-02.mp3)
+
+**English 3**
+
+<audio controls preload="none" src="assets/showcase/english-agentic-03.mp3"></audio>
+
+[Download MP3](assets/showcase/english-agentic-03.mp3)
+
+## Model architecture
+
+| Predictor | Role | Transformer blocks | Width | Stored parameters |
+|---|---|---:|---:|---:|
+| CB0 | Semantic audio and duration | 28 | 2,048 | 1,829,116,930 |
+| CB1 | Acoustic refinement | 16 | 1,024 | 448,960,512 |
+| CB2 | Acoustic refinement | 8 | 1,024 | 327,307,264 |
+| CB3 | Acoustic refinement | 4 | 1,024 | 268,577,792 |
+| **Total** |  | **56** |  | **2,873,962,498** |
+
+CB0 uses a Qwen3-1.7B-derived transformer trunk. CB1--CB3 use progressively
+shallower Qwen3-0.6B-derived trunks. Each predictor replaces the inherited
+language-model output layer with a two-layer audio-token head. The four output
+streams are decoded with DualCodec; the release inference path then uses the
+acoustic encoder and decoder from VibeVoice for continuous reconstruction and
+streaming.
+
+## Training
+
+All four predictors were trained exclusively with supervised fine-tuning on
+approximately 200,000 hours of paired speech and text across seven languages,
+predominantly from public-domain audiobook recordings and openly released
+speech corpora.
+
+## Repository contents
+
+```text
+TontaubeV1/
+├── cb0/
+├── cb1/
+├── cb2/
+└── cb3/
+```
+
+Each directory contains its safetensors weights, runtime configuration, and
+tokenizer files. The repository does not redistribute DualCodec, W2V-BERT, or
+VibeVoice weights. The inference setup downloads those components from their
+upstream repositories at pinned revisions.
+
+## Serving performance
+
+With weights resident on one NVIDIA GeForce RTX 5090 and the process warmed,
+the streaming path reaches approximately 200 ms to first encoded audio. In
+separate non-streaming measurements, end-to-end real-time factor (RTF) is 0.08
+for one input and aggregate RTF is 0.02 across eight concurrent inputs. Startup,
+model loading, and network latency are excluded.
+
+## Evaluation
+
+The LLM-as-a-judge pairwise audiobook-reading benchmark contains 400 fixed
+English passages of 250--500 characters. For each passage, Gemini 3.1 Pro
+Preview judges the same output pair twice, once in each presentation order, on
+prosody and word-by-word correctness. A TontaubeV1 preference, tie, or
+comparator preference scores 1, 0.5, or 0; the figure reports the mean over all
+800 order-balanced judgments. A score of 50% denotes parity. Whiskers show 95%
+passage-cluster bootstrap intervals; labels beneath the points give tie rates
+across individual judge calls.
+
+<p align="center">
+  <img src="assets/llm-judge-results.png" alt="TontaubeV1 preference scores against ElevenLabs Flash v2.5, Fish Audio S2 Pro, Gradium, and Cartesia on prosody and correctness.">
+</p>
+
+<details>
+<summary>Exact preference scores</summary>
+
+| Comparator | Prosody preference | Correctness preference |
+|---|---:|---:|
+| ElevenLabs Flash v2.5 | 50.1% | 48.9% |
+| Fish Audio S2 Pro | 82.1% | 49.6% |
+| Gradium API, April 2026 | 86.2% | 54.6% |
+| Cartesia Sonic 3 | 82.3% | 60.8% |
+
+</details>
+
+TontaubeV1 uses semantic sampling temperature `0.55` and acoustic temperature
+zero in these comparisons. Each waveform is independently normalized to an
+average level of -20 dBFS before judging. The technical report gives the
+complete judge instructions and further methodological limitations.
+
+Fish Audio uses the same frozen cloning reference as TontaubeV1. ElevenLabs,
+Gradium, and Cartesia instead use fixed provider voices while TontaubeV1 uses
+the cloning reference. Although voice identity and timbre are excluded from the
+rubric, prosody is not fully separable from the reference; this asymmetry may
+favor TontaubeV1 in those three comparisons.
+
+On the 1,088 English zero-shot examples of the Seed-TTS evaluation set,
+TontaubeV1 obtains 1.66% mean utterance-level word error rate using Whisper
+large-v3 transcription at semantic sampling temperature `0.6`.
+
+These automated evaluations measure English reading prosody and text
+correctness. They do not establish voice similarity, overall sound quality,
+multilingual quality, long-form continuity, or streaming quality.
+
+## Limitations and responsible use
+
+Autoregressive generation can omit, repeat, or alter text and can terminate
+too early or too late. Reference conditioning may reproduce incidental
+recording properties or imperfectly preserve identity. Quality varies by
+language, speaker, style, and text domain. The optional verbalizer can
+normalize incorrectly or alter wording. Training emphasized audiobook speech,
+so audiobook generation may be more reliable than conversational or agentic
+generation.
+
+English and German are the primary languages. German phoneme realization can
+be imperfect, and Spanish, French, Italian, Dutch, and Portuguese have not been
+evaluated by native speakers; no comparative quality claims are made for those
+languages.
+
+Voice cloning can enable impersonation, fraud, non-consensual synthesis, and
+misleading media. Obtain permission for reference voices and apply appropriate
+authentication, rate limiting, disclosure, logging, and abuse-response
+measures in deployed systems. Generated speech must not be treated as verified
+identity or factual evidence.
+
+## License and attribution
+
+TontaubeV1 is distributed under the
+[Tontaube Community Model License 1.0](LICENSE), which is not an open-source
+license. The license applies to the Tontaube model materials identified by the
+release manifest; third-party material remains subject to its respective
+terms. See [Third-Party Notices](THIRD_PARTY_NOTICES.md).
+
+TontaubeV1 builds on Qwen3 and uses the DualCodec speech representation. The
+release inference stack also uses components from VibeVoice, W2V-BERT,
+PyTorch, Transformers, and vLLM. Their respective licenses and notices remain
+applicable.
+
+The generated audio examples in `assets/showcase/` are provided as
+demonstration material and are not part of the model materials covered by the
+Tontaube Community Model License 1.0.

THIRD_PARTY_LICENSES/QWEN3-APACHE-2.0.txt

@@ -0,0 +1,201 @@
+                                 Apache License
+                           Version 2.0, January 2004
+                        http://www.apache.org/licenses/
+
+   TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
+
+   1. Definitions.
+
+      "License" shall mean the terms and conditions for use, reproduction,
+      and distribution as defined by Sections 1 through 9 of this document.
+
+      "Licensor" shall mean the copyright owner or entity authorized by
+      the copyright owner that is granting the License.
+
+      "Legal Entity" shall mean the union of the acting entity and all
+      other entities that control, are controlled by, or are under common
+      control with that entity. For the purposes of this definition,
+      "control" means (i) the power, direct or indirect, to cause the
+      direction or management of such entity, whether by contract or
+      otherwise, or (ii) ownership of fifty percent (50%) or more of the
+      outstanding shares, or (iii) beneficial ownership of such entity.
+
+      "You" (or "Your") shall mean an individual or Legal Entity
+      exercising permissions granted by this License.
+
+      "Source" form shall mean the preferred form for making modifications,
+      including but not limited to software source code, documentation
+      source, and configuration files.
+
+      "Object" form shall mean any form resulting from mechanical
+      transformation or translation of a Source form, including but
+      not limited to compiled object code, generated documentation,
+      and conversions to other media types.
+
+      "Work" shall mean the work of authorship, whether in Source or
+      Object form, made available under the License, as indicated by a
+      copyright notice that is included in or attached to the work
+      (an example is provided in the Appendix below).
+
+      "Derivative Works" shall mean any work, whether in Source or Object
+      form, that is based on (or derived from) the Work and for which the
+      editorial revisions, annotations, elaborations, or other modifications
+      represent, as a whole, an original work of authorship. For the purposes
+      of this License, Derivative Works shall not include works that remain
+      separable from, or merely link (or bind by name) to the interfaces of,
+      the Work and Derivative Works thereof.
+
+      "Contribution" shall mean any work of authorship, including
+      the original version of the Work and any modifications or additions
+      to that Work or Derivative Works thereof, that is intentionally
+      submitted to Licensor for inclusion in the Work by the copyright owner
+      or by an individual or Legal Entity authorized to submit on behalf of
+      the copyright owner. For the purposes of this definition, "submitted"
+      means any form of electronic, verbal, or written communication sent
+      to the Licensor or its representatives, including but not limited to
+      communication on electronic mailing lists, source code control systems,
+      and issue tracking systems that are managed by, or on behalf of, the
+      Licensor for the purpose of discussing and improving the Work, but
+      excluding communication that is conspicuously marked or otherwise
+      designated in writing by the copyright owner as "Not a Contribution."
+
+      "Contributor" shall mean Licensor and any individual or Legal Entity
+      on behalf of whom a Contribution has been received by Licensor and
+      subsequently incorporated within the Work.
+
+   2. Grant of Copyright License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      copyright license to reproduce, prepare Derivative Works of,
+      publicly display, publicly perform, sublicense, and distribute the
+      Work and such Derivative Works in Source or Object form.
+
+   3. Grant of Patent License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      (except as stated in this section) patent license to make, have made,
+      use, offer to sell, sell, import, and otherwise transfer the Work,
+      where such license applies only to those patent claims licensable
+      by such Contributor that are necessarily infringed by their
+      Contribution(s) alone or by combination of their Contribution(s)
+      with the Work to which such Contribution(s) was submitted. If You
+      institute patent litigation against any entity (including a
+      cross-claim or counterclaim in a lawsuit) alleging that the Work
+      or a Contribution incorporated within the Work constitutes direct
+      or contributory patent infringement, then any patent licenses
+      granted to You under this License for that Work shall terminate
+      as of the date such litigation is filed.
+
+   4. Redistribution. You may reproduce and distribute copies of the
+      Work or Derivative Works thereof in any medium, with or without
+      modifications, and in Source or Object form, provided that You
+      meet the following conditions:
+
+      (a) You must give any other recipients of the Work or
+          Derivative Works a copy of this License; and
+
+      (b) You must cause any modified files to carry prominent notices
+          stating that You changed the files; and
+
+      (c) You must retain, in the Source form of any Derivative Works
+          that You distribute, all copyright, patent, trademark, and
+          attribution notices from the Source form of the Work,
+          excluding those notices that do not pertain to any part of
+          the Derivative Works; and
+
+      (d) If the Work includes a "NOTICE" text file as part of its
+          distribution, then any Derivative Works that You distribute must
+          include a readable copy of the attribution notices contained
+          within such NOTICE file, excluding those notices that do not
+          pertain to any part of the Derivative Works, in at least one
+          of the following places: within a NOTICE text file distributed
+          as part of the Derivative Works; within the Source form or
+          documentation, if provided along with the Derivative Works; or,
+          within a display generated by the Derivative Works, if and
+          wherever such third-party notices normally appear. The contents
+          of the NOTICE file are for informational purposes only and
+          do not modify the License. You may add Your own attribution
+          notices within Derivative Works that You distribute, alongside
+          or as an addendum to the NOTICE text from the Work, provided
+          that such additional attribution notices cannot be construed
+          as modifying the License.
+
+      You may add Your own copyright statement to Your modifications and
+      may provide additional or different license terms and conditions
+      for use, reproduction, or distribution of Your modifications, or
+      for any such Derivative Works as a whole, provided Your use,
+      reproduction, and distribution of the Work otherwise complies with
+      the conditions stated in this License.
+
+   5. Submission of Contributions. Unless You explicitly state otherwise,
+      any Contribution intentionally submitted for inclusion in the Work
+      by You to the Licensor shall be under the terms and conditions of
+      this License, without any additional terms or conditions.
+      Notwithstanding the above, nothing herein shall supersede or modify
+      the terms of any separate license agreement you may have executed
+      with Licensor regarding such Contributions.
+
+   6. Trademarks. This License does not grant permission to use the trade
+      names, trademarks, service marks, or product names of the Licensor,
+      except as required for reasonable and customary use in describing the
+      origin of the Work and reproducing the content of the NOTICE file.
+
+   7. Disclaimer of Warranty. Unless required by applicable law or
+      agreed to in writing, Licensor provides the Work (and each
+      Contributor provides its Contributions) on an "AS IS" BASIS,
+      WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
+      implied, including, without limitation, any warranties or conditions
+      of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
+      PARTICULAR PURPOSE. You are solely responsible for determining the
+      appropriateness of using or redistributing the Work and assume any
+      risks associated with Your exercise of permissions under this License.
+
+   8. Limitation of Liability. In no event and under no legal theory,
+      whether in tort (including negligence), contract, or otherwise,
+      unless required by applicable law (such as deliberate and grossly
+      negligent acts) or agreed to in writing, shall any Contributor be
+      liable to You for damages, including any direct, indirect, special,
+      incidental, or consequential damages of any character arising as a
+      result of this License or out of the use or inability to use the
+      Work (including but not limited to damages for loss of goodwill,
+      work stoppage, computer failure or malfunction, or any and all
+      other commercial damages or losses), even if such Contributor
+      has been advised of the possibility of such damages.
+
+   9. Accepting Warranty or Additional Liability. While redistributing
+      the Work or Derivative Works thereof, You may choose to offer,
+      and charge a fee for, acceptance of support, warranty, indemnity,
+      or other liability obligations and/or rights consistent with this
+      License. However, in accepting such obligations, You may act only
+      on Your own behalf and on Your sole responsibility, not on behalf
+      of any other Contributor, and only if You agree to indemnify,
+      defend, and hold each Contributor harmless for any liability
+      incurred by, or claims asserted against, such Contributor by reason
+      of your accepting any such warranty or additional liability.
+
+   END OF TERMS AND CONDITIONS
+
+   APPENDIX: How to apply the Apache License to your work.
+
+      To apply the Apache License to your work, attach the following
+      boilerplate notice, with the fields enclosed by brackets "[]"
+      replaced with your own identifying information. (Don't include
+      the brackets!)  The text should be enclosed in the appropriate
+      comment syntax for the file format. We also recommend that a
+      file or class name and description of purpose be included on the
+      same "printed page" as the copyright notice for easier
+      identification within third-party archives.
+
+   Copyright 2024 Alibaba Cloud
+
+   Licensed under the Apache License, Version 2.0 (the "License");
+   you may not use this file except in compliance with the License.
+   You may obtain a copy of the License at
+
+       http://www.apache.org/licenses/LICENSE-2.0
+
+   Unless required by applicable law or agreed to in writing, software
+   distributed under the License is distributed on an "AS IS" BASIS,
+   WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+   See the License for the specific language governing permissions and
+   limitations under the License.

THIRD_PARTY_NOTICES.md

@@ -0,0 +1,25 @@
+# Third-Party Notices
+
+This file identifies third-party material incorporated into the TontaubeV1
+model artifacts. It does not replace or modify any applicable license.
+
+## Qwen3
+
+TontaubeV1 is derived from the following Qwen3 models:
+
+- CB0: [Qwen3-1.7B](https://huggingface.co/Qwen/Qwen3-1.7B)
+- CB1, CB2, and CB3: [Qwen3-0.6B](https://huggingface.co/Qwen/Qwen3-0.6B)
+
+Qwen3 is copyright 2024 Alibaba Cloud and is distributed under the Apache
+License 2.0. A copy of that license is provided in
+[`THIRD_PARTY_LICENSES/QWEN3-APACHE-2.0.txt`](THIRD_PARTY_LICENSES/QWEN3-APACHE-2.0.txt).
+
+The Tontaube Community Model License 1.0 applies only to rights owned by or
+licensable by Cremer & Cremer Technologies UG (haftungsbeschränkt) in the
+files identified by [`COVERED_FILES.md`](COVERED_FILES.md). It does not replace
+the Apache License 2.0 terms applicable to Qwen3 material incorporated into
+those files.
+
+DualCodec, Wav2Vec-BERT, and VibeVoice weights are not included in this model
+repository. They are obtained separately by the inference implementation and
+remain subject to their own licenses and notices.

assets/showcase/english-agentic-01.mp3

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:016143c3880d2f06f4bc66c53ef6a9f3e0085bfdd9060841c86bf3ea87efe0ac
+size 167936

assets/showcase/english-agentic-02.mp3

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7bd14b351b74cfc5add7d6035d1b9255cf61abe5e7c72a6f3498f665578aad89
+size 146434

assets/showcase/english-agentic-03.mp3

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9ce0122f7bf2abfb047c5f05b7d15686372c7ea4241e80d1aa277f33b483b926
+size 161408

assets/showcase/english-audiobook-01.mp3

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ef45059033f48282b709696923552ce7de152a0529d972369b219f3e6f72e945
+size 278924

assets/showcase/english-audiobook-02.mp3

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e8b9ac5eb26b3a2505f8c5ed61a666402c01235b052ad63a65dce7ea17ca768a
+size 430176

assets/showcase/english-audiobook-03.mp3

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:987e6082ac50ee123c8d8496bfb6113c329f045a3205acfb911b308db2646560
+size 244704

assets/showcase/german-audiobook-01.mp3

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bfe1bba89876f0986fa541ef77df5935ee175854e6c30aaad3a8debf4771f951
+size 422879

cb0/config.json

@@ -0,0 +1,97 @@
+{
+  "vocab_size": 168057,
+  "max_position_embeddings": 40960,
+  "hidden_size": 2048,
+  "intermediate_size": 6144,
+  "num_hidden_layers": 28,
+  "num_attention_heads": 16,
+  "use_sliding_window": false,
+  "sliding_window": null,
+  "max_window_layers": 28,
+  "num_key_value_heads": 8,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "initializer_range": 0.02,
+  "rms_norm_eps": 1e-06,
+  "use_cache": true,
+  "rope_theta": 1000000,
+  "rope_scaling": null,
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "return_dict": true,
+  "output_hidden_states": false,
+  "torchscript": false,
+  "dtype": "bfloat16",
+  "pruned_heads": {},
+  "tie_word_embeddings": false,
+  "chunk_size_feed_forward": 0,
+  "is_encoder_decoder": false,
+  "is_decoder": false,
+  "cross_attention_hidden_size": null,
+  "add_cross_attention": false,
+  "tie_encoder_decoder": false,
+  "architectures": [
+    "Qwen3DualCodecForCausalLM"
+  ],
+  "finetuning_task": null,
+  "id2label": {
+    "0": "LABEL_0",
+    "1": "LABEL_1"
+  },
+  "label2id": {
+    "LABEL_0": 0,
+    "LABEL_1": 1
+  },
+  "task_specific_params": null,
+  "problem_type": null,
+  "tokenizer_class": null,
+  "prefix": null,
+  "bos_token_id": 151643,
+  "pad_token_id": 151643,
+  "eos_token_id": 151670,
+  "sep_token_id": null,
+  "decoder_start_token_id": null,
+  "transformers_version": "4.57.6",
+  "model_type": "qwen3",
+  "tf_legacy_loss": false,
+  "use_bfloat16": false,
+  "output_attentions": false,
+  "audio_token_start_id": 151673,
+  "n_codebooks": 4,
+  "target_codebook": 0,
+  "audio_head_hidden": 4096,
+  "text_marker_token_id": 151671,
+  "audio_marker_token_id": 151672,
+  "chunk_realign_offset": 25,
+  "_vllm_config_version": 6
+}

cb0/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:17f990196944ada19734e43c195eaf998c998173edba00d14f01a4668865e814
+size 3809322216

cb0/tokenizer.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8c838573f03edfd45dce6890c0ed9cb9ea51a3721f7385f6072c8bd2919d37ed
+size 16873051

cb1/config.json

@@ -0,0 +1,85 @@
+{
+  "vocab_size": 172153,
+  "max_position_embeddings": 40960,
+  "hidden_size": 1024,
+  "intermediate_size": 3072,
+  "num_hidden_layers": 16,
+  "num_attention_heads": 16,
+  "use_sliding_window": false,
+  "sliding_window": null,
+  "max_window_layers": 28,
+  "num_key_value_heads": 8,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "initializer_range": 0.02,
+  "rms_norm_eps": 1e-06,
+  "use_cache": true,
+  "rope_theta": 1000000,
+  "rope_scaling": null,
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "return_dict": true,
+  "output_hidden_states": false,
+  "torchscript": false,
+  "dtype": "bfloat16",
+  "pruned_heads": {},
+  "tie_word_embeddings": false,
+  "chunk_size_feed_forward": 0,
+  "is_encoder_decoder": false,
+  "is_decoder": false,
+  "cross_attention_hidden_size": null,
+  "add_cross_attention": false,
+  "tie_encoder_decoder": false,
+  "architectures": [
+    "Qwen3DualCodecForCausalLM"
+  ],
+  "finetuning_task": null,
+  "id2label": {
+    "0": "LABEL_0",
+    "1": "LABEL_1"
+  },
+  "label2id": {
+    "LABEL_0": 0,
+    "LABEL_1": 1
+  },
+  "task_specific_params": null,
+  "problem_type": null,
+  "tokenizer_class": null,
+  "prefix": null,
+  "bos_token_id": 151643,
+  "pad_token_id": 151643,
+  "eos_token_id": 151670,
+  "sep_token_id": null,
+  "decoder_start_token_id": null,
+  "transformers_version": "4.57.6",
+  "model_type": "qwen3",
+  "tf_legacy_loss": false,
+  "use_bfloat16": false,
+  "output_attentions": false,
+  "audio_token_start_id": 151673,
+  "n_codebooks": 4,
+  "target_codebook": 1,
+  "audio_head_hidden": 4096,
+  "text_marker_token_id": 151671,
+  "audio_marker_token_id": 151672,
+  "chunk_realign_offset": 25,
+  "_vllm_config_version": 6
+}

cb1/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a2404b2b1f0ab6e92e82047e06d273821f3b413fedb8ad27b234d5526bf281c1
+size 939900896

cb1/tokenizer.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8c838573f03edfd45dce6890c0ed9cb9ea51a3721f7385f6072c8bd2919d37ed
+size 16873051

cb2/config.json

@@ -0,0 +1,77 @@
+{
+  "vocab_size": 176249,
+  "max_position_embeddings": 40960,
+  "hidden_size": 1024,
+  "intermediate_size": 3072,
+  "num_hidden_layers": 8,
+  "num_attention_heads": 16,
+  "use_sliding_window": false,
+  "sliding_window": null,
+  "max_window_layers": 28,
+  "num_key_value_heads": 8,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "initializer_range": 0.02,
+  "rms_norm_eps": 1e-06,
+  "use_cache": true,
+  "rope_theta": 1000000,
+  "rope_scaling": null,
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "return_dict": true,
+  "output_hidden_states": false,
+  "torchscript": false,
+  "dtype": "bfloat16",
+  "pruned_heads": {},
+  "tie_word_embeddings": false,
+  "chunk_size_feed_forward": 0,
+  "is_encoder_decoder": false,
+  "is_decoder": false,
+  "cross_attention_hidden_size": null,
+  "add_cross_attention": false,
+  "tie_encoder_decoder": false,
+  "architectures": [
+    "Qwen3DualCodecForCausalLM"
+  ],
+  "finetuning_task": null,
+  "id2label": {
+    "0": "LABEL_0",
+    "1": "LABEL_1"
+  },
+  "label2id": {
+    "LABEL_0": 0,
+    "LABEL_1": 1
+  },
+  "task_specific_params": null,
+  "problem_type": null,
+  "tokenizer_class": null,
+  "prefix": null,
+  "bos_token_id": 151643,
+  "pad_token_id": 151643,
+  "eos_token_id": 151670,
+  "sep_token_id": null,
+  "decoder_start_token_id": null,
+  "transformers_version": "4.57.6",
+  "model_type": "qwen3",
+  "tf_legacy_loss": false,
+  "use_bfloat16": false,
+  "output_attentions": false,
+  "audio_token_start_id": 151673,
+  "n_codebooks": 4,
+  "target_codebook": 2,
+  "audio_head_hidden": 4096,
+  "text_marker_token_id": 151671,
+  "audio_marker_token_id": 151672,
+  "chunk_realign_offset": 25,
+  "_vllm_config_version": 6
+}

cb2/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:78fd0709da355a64438505e726ecaa49ecfb41c4a096298d8b88fd7566baa081
+size 696584424

cb2/tokenizer.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8c838573f03edfd45dce6890c0ed9cb9ea51a3721f7385f6072c8bd2919d37ed
+size 16873051

cb3/config.json

@@ -0,0 +1,73 @@
+{
+  "vocab_size": 180345,
+  "max_position_embeddings": 40960,
+  "hidden_size": 1024,
+  "intermediate_size": 3072,
+  "num_hidden_layers": 4,
+  "num_attention_heads": 16,
+  "use_sliding_window": false,
+  "sliding_window": null,
+  "max_window_layers": 28,
+  "num_key_value_heads": 8,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "initializer_range": 0.02,
+  "rms_norm_eps": 1e-06,
+  "use_cache": true,
+  "rope_theta": 1000000,
+  "rope_scaling": null,
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "return_dict": true,
+  "output_hidden_states": false,
+  "torchscript": false,
+  "dtype": "bfloat16",
+  "pruned_heads": {},
+  "tie_word_embeddings": false,
+  "chunk_size_feed_forward": 0,
+  "is_encoder_decoder": false,
+  "is_decoder": false,
+  "cross_attention_hidden_size": null,
+  "add_cross_attention": false,
+  "tie_encoder_decoder": false,
+  "architectures": [
+    "Qwen3DualCodecForCausalLM"
+  ],
+  "finetuning_task": null,
+  "id2label": {
+    "0": "LABEL_0",
+    "1": "LABEL_1"
+  },
+  "label2id": {
+    "LABEL_0": 0,
+    "LABEL_1": 1
+  },
+  "task_specific_params": null,
+  "problem_type": null,
+  "tokenizer_class": null,
+  "prefix": null,
+  "bos_token_id": 151643,
+  "pad_token_id": 151643,
+  "eos_token_id": 151670,
+  "sep_token_id": null,
+  "decoder_start_token_id": null,
+  "transformers_version": "4.57.6",
+  "model_type": "qwen3",
+  "tf_legacy_loss": false,
+  "use_bfloat16": false,
+  "output_attentions": false,
+  "audio_token_start_id": 151673,
+  "n_codebooks": 4,
+  "target_codebook": 3,
+  "audio_head_hidden": 4096,
+  "text_marker_token_id": 151671,
+  "audio_marker_token_id": 151672,
+  "chunk_realign_offset": 25,
+  "_vllm_config_version": 6
+}

cb3/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:09dab1fb894af2cd98ff8a2bf784700a78ca67be749144fb02a7feb947d779bf
+size 579120520

cb3/tokenizer.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8c838573f03edfd45dce6890c0ed9cb9ea51a3721f7385f6072c8bd2919d37ed
+size 16873051