Add files using upload-large-folder tool
Browse files- OmniVoice_AudioEncoder.mlpackage/Data/com.apple.CoreML/model.mlmodel +3 -0
- OmniVoice_AudioEncoder.mlpackage/Data/com.apple.CoreML/weights/weight.bin +3 -0
- OmniVoice_AudioEncoder.mlpackage/Manifest.json +18 -0
- OmniVoice_AudioEncoder.mlpackage/executorch_debug_handle_mapping.json +0 -0
- OmniVoice_GeneratorStep.mlpackage/Data/com.apple.CoreML/model.mlmodel +3 -0
- OmniVoice_GeneratorStep.mlpackage/Data/com.apple.CoreML/weights/weight.bin +3 -0
- OmniVoice_GeneratorStep.mlpackage/Manifest.json +18 -0
- OmniVoice_GeneratorStep.mlpackage/executorch_debug_handle_mapping.json +0 -0
- OmniVoice_LM.mlpackage/Data/com.apple.CoreML/model.mlmodel +3 -0
- OmniVoice_LM.mlpackage/Data/com.apple.CoreML/weights/weight.bin +3 -0
- OmniVoice_LM.mlpackage/Manifest.json +18 -0
- OmniVoice_SpeakerEncoder.mlpackage/Data/com.apple.CoreML/model.mlmodel +3 -0
- OmniVoice_SpeakerEncoder.mlpackage/Data/com.apple.CoreML/weights/weight.bin +3 -0
- OmniVoice_SpeakerEncoder.mlpackage/Manifest.json +18 -0
- OmniVoice_Vocoder.mlpackage/Data/com.apple.CoreML/model.mlmodel +3 -0
- OmniVoice_Vocoder.mlpackage/Data/com.apple.CoreML/weights/weight.bin +3 -0
- OmniVoice_Vocoder.mlpackage/Manifest.json +18 -0
- README.md +18 -0
- REAL_GENERATOR_README.txt +2 -0
- STUB_README.txt +1 -0
- tokens.txt +0 -0
OmniVoice_AudioEncoder.mlpackage/Data/com.apple.CoreML/model.mlmodel
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6b4fb24c9ecee1e14045c418a891cadddb881e8e98a4f04baf39662d4f6a05c2
|
| 3 |
+
size 343923
|
OmniVoice_AudioEncoder.mlpackage/Data/com.apple.CoreML/weights/weight.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3f0b2263b71d41e8d6e7514942490e878fccb84f46d42b13d789852442e83744
|
| 3 |
+
size 326076352
|
OmniVoice_AudioEncoder.mlpackage/Manifest.json
ADDED
|
@@ -0,0 +1,18 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"fileFormatVersion": "1.0.0",
|
| 3 |
+
"itemInfoEntries": {
|
| 4 |
+
"7D8F84C6-E6FD-4795-BA97-5BDB19CB0BD4": {
|
| 5 |
+
"author": "com.apple.CoreML",
|
| 6 |
+
"description": "CoreML Model Weights",
|
| 7 |
+
"name": "weights",
|
| 8 |
+
"path": "com.apple.CoreML/weights"
|
| 9 |
+
},
|
| 10 |
+
"DCA64EBA-A273-4D29-B732-660DAB49BC50": {
|
| 11 |
+
"author": "com.apple.CoreML",
|
| 12 |
+
"description": "CoreML Model Specification",
|
| 13 |
+
"name": "model.mlmodel",
|
| 14 |
+
"path": "com.apple.CoreML/model.mlmodel"
|
| 15 |
+
}
|
| 16 |
+
},
|
| 17 |
+
"rootModelIdentifier": "DCA64EBA-A273-4D29-B732-660DAB49BC50"
|
| 18 |
+
}
|
OmniVoice_AudioEncoder.mlpackage/executorch_debug_handle_mapping.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
OmniVoice_GeneratorStep.mlpackage/Data/com.apple.CoreML/model.mlmodel
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1c9f58782cf2ea4a73689aad380ce08f3c8ce8fdacb8033888dbd190cb053d67
|
| 3 |
+
size 644271
|
OmniVoice_GeneratorStep.mlpackage/Data/com.apple.CoreML/weights/weight.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7944334a4047459b144d30c9f6fad572227d2d98ef30227974eda91c1b205ce0
|
| 3 |
+
size 1225220048
|
OmniVoice_GeneratorStep.mlpackage/Manifest.json
ADDED
|
@@ -0,0 +1,18 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"fileFormatVersion": "1.0.0",
|
| 3 |
+
"itemInfoEntries": {
|
| 4 |
+
"7BE607D8-6C1C-4003-8BB3-54E19B7AAB76": {
|
| 5 |
+
"author": "com.apple.CoreML",
|
| 6 |
+
"description": "CoreML Model Weights",
|
| 7 |
+
"name": "weights",
|
| 8 |
+
"path": "com.apple.CoreML/weights"
|
| 9 |
+
},
|
| 10 |
+
"BC72B755-42F2-4EC2-9AFB-20BB9C94A614": {
|
| 11 |
+
"author": "com.apple.CoreML",
|
| 12 |
+
"description": "CoreML Model Specification",
|
| 13 |
+
"name": "model.mlmodel",
|
| 14 |
+
"path": "com.apple.CoreML/model.mlmodel"
|
| 15 |
+
}
|
| 16 |
+
},
|
| 17 |
+
"rootModelIdentifier": "BC72B755-42F2-4EC2-9AFB-20BB9C94A614"
|
| 18 |
+
}
|
OmniVoice_GeneratorStep.mlpackage/executorch_debug_handle_mapping.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
OmniVoice_LM.mlpackage/Data/com.apple.CoreML/model.mlmodel
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e69d4f8d783d7f644634a627c97b4c5f48d59dcf2cff310c6d1799182b4b4ea5
|
| 3 |
+
size 3375
|
OmniVoice_LM.mlpackage/Data/com.apple.CoreML/weights/weight.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:41d30251363e9f0382602d2f45138db1bbfdd14e73476366972121aebb5b4d87
|
| 3 |
+
size 2099392
|
OmniVoice_LM.mlpackage/Manifest.json
ADDED
|
@@ -0,0 +1,18 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"fileFormatVersion": "1.0.0",
|
| 3 |
+
"itemInfoEntries": {
|
| 4 |
+
"860E840A-07FB-4CC2-BC59-DB644ACE74DD": {
|
| 5 |
+
"author": "com.apple.CoreML",
|
| 6 |
+
"description": "CoreML Model Specification",
|
| 7 |
+
"name": "model.mlmodel",
|
| 8 |
+
"path": "com.apple.CoreML/model.mlmodel"
|
| 9 |
+
},
|
| 10 |
+
"C44E96E3-967F-45BD-87DE-DE0F37C634A9": {
|
| 11 |
+
"author": "com.apple.CoreML",
|
| 12 |
+
"description": "CoreML Model Weights",
|
| 13 |
+
"name": "weights",
|
| 14 |
+
"path": "com.apple.CoreML/weights"
|
| 15 |
+
}
|
| 16 |
+
},
|
| 17 |
+
"rootModelIdentifier": "860E840A-07FB-4CC2-BC59-DB644ACE74DD"
|
| 18 |
+
}
|
OmniVoice_SpeakerEncoder.mlpackage/Data/com.apple.CoreML/model.mlmodel
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:67f6de1867048d5c17bc402fa94a95307ce9bbb33b974b17f3dab23f71384c9a
|
| 3 |
+
size 2122
|
OmniVoice_SpeakerEncoder.mlpackage/Data/com.apple.CoreML/weights/weight.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ed223fc704b76f5ddb09cc9e764a582626fd60204486bb76ea970b0fb7f11f43
|
| 3 |
+
size 32770240
|
OmniVoice_SpeakerEncoder.mlpackage/Manifest.json
ADDED
|
@@ -0,0 +1,18 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"fileFormatVersion": "1.0.0",
|
| 3 |
+
"itemInfoEntries": {
|
| 4 |
+
"A1FA9F94-CA4B-4A83-950E-EF9603F80450": {
|
| 5 |
+
"author": "com.apple.CoreML",
|
| 6 |
+
"description": "CoreML Model Specification",
|
| 7 |
+
"name": "model.mlmodel",
|
| 8 |
+
"path": "com.apple.CoreML/model.mlmodel"
|
| 9 |
+
},
|
| 10 |
+
"A840898E-5F89-43BB-B999-7E97F2A41D00": {
|
| 11 |
+
"author": "com.apple.CoreML",
|
| 12 |
+
"description": "CoreML Model Weights",
|
| 13 |
+
"name": "weights",
|
| 14 |
+
"path": "com.apple.CoreML/weights"
|
| 15 |
+
}
|
| 16 |
+
},
|
| 17 |
+
"rootModelIdentifier": "A1FA9F94-CA4B-4A83-950E-EF9603F80450"
|
| 18 |
+
}
|
OmniVoice_Vocoder.mlpackage/Data/com.apple.CoreML/model.mlmodel
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b737ab5ffba7f4464a426b7df0a7410f392999e1ff544de71ac5210cd585efa3
|
| 3 |
+
size 142482
|
OmniVoice_Vocoder.mlpackage/Data/com.apple.CoreML/weights/weight.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8be8b529c7e4ddfbbccddad3a3aef5014b42fa01a2bfd853a0af3d76a15dce02
|
| 3 |
+
size 43145344
|
OmniVoice_Vocoder.mlpackage/Manifest.json
ADDED
|
@@ -0,0 +1,18 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"fileFormatVersion": "1.0.0",
|
| 3 |
+
"itemInfoEntries": {
|
| 4 |
+
"9F7DC86C-44B4-4210-91D1-0075D343C310": {
|
| 5 |
+
"author": "com.apple.CoreML",
|
| 6 |
+
"description": "CoreML Model Weights",
|
| 7 |
+
"name": "weights",
|
| 8 |
+
"path": "com.apple.CoreML/weights"
|
| 9 |
+
},
|
| 10 |
+
"C1CDFF82-CDF7-4691-9789-3767F2CED598": {
|
| 11 |
+
"author": "com.apple.CoreML",
|
| 12 |
+
"description": "CoreML Model Specification",
|
| 13 |
+
"name": "model.mlmodel",
|
| 14 |
+
"path": "com.apple.CoreML/model.mlmodel"
|
| 15 |
+
}
|
| 16 |
+
},
|
| 17 |
+
"rootModelIdentifier": "C1CDFF82-CDF7-4691-9789-3767F2CED598"
|
| 18 |
+
}
|
README.md
ADDED
|
@@ -0,0 +1,18 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# OmniVoice Core ML bundles (TranslateBlue export)
|
| 2 |
+
|
| 3 |
+
Private artifact drop for **on-device** experiments. Upstream weights: [`k2-fsa/OmniVoice`](https://huggingface.co/k2-fsa/OmniVoice).
|
| 4 |
+
|
| 5 |
+
## Contents
|
| 6 |
+
|
| 7 |
+
| Path | Description |
|
| 8 |
+
|------|-------------|
|
| 9 |
+
| `OmniVoice_Vocoder.mlpackage` | Higgs **decode** (discrete codes → waveform). |
|
| 10 |
+
| `OmniVoice_GeneratorStep.mlpackage` | **Real** `OmniVoice.forward` logits (`torch.export` + coremltools). |
|
| 11 |
+
| `OmniVoice_AudioEncoder.mlpackage` | **Real** Higgs **encode** (waveform → codes). |
|
| 12 |
+
| `OmniVoice_LM.mlpackage` | **Stub** I/O only (not trained weights). |
|
| 13 |
+
| `OmniVoice_SpeakerEncoder.mlpackage` | **Stub** placeholder. |
|
| 14 |
+
| `tokens.txt` | Text tokenizer vocabulary for Swift. |
|
| 15 |
+
|
| 16 |
+
Export scripts live in the TranslateBlue repo under `Scripts/export_omnivoice_coreml/`. See `Docs/OmniVoice_CoreML_Contract.md` for Swift vs Python I/O notes.
|
| 17 |
+
|
| 18 |
+
**License:** Follow the license terms of `k2-fsa/OmniVoice` and third-party components (Qwen3, Higgs, etc.). Do not redistribute if your use case is not permitted by those licenses.
|
REAL_GENERATOR_README.txt
ADDED
|
@@ -0,0 +1,2 @@
|
|
|
|
|
|
|
|
|
|
| 1 |
+
OmniVoice_GeneratorStep.mlpackage is the real OmniVoice.forward logits export (C=8, V=1025, max_seq=32, batch rows=2 for B=1 CFG layout).
|
| 2 |
+
Swift DiffusionEngine still expects OmniVoice_LM.mlpackage stub I/O until you implement upstream masked decoding.
|
STUB_README.txt
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
OmniVoice_LM.mlpackage was built with export_lm_step.py --stub (NOT trained OmniVoice).
|
tokens.txt
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|