🚀 F5-TTSx: A Fairytaler that Fakes Fluent and Faithful Speech with Flow Matching eXtended
🔧 About
F5-TTSx is a custom build of the F5-TTS voice synthezer, designed to get more fliud settings for synthezing voice - especially on systems with limited VRAM.
🔧 Key Differences from Official F5-TTS
- Optimized for all systems with nVidia GPUs
- Main
.batmenu with option to install/reinstall project - Enhanced UI settings with useful features and stability improvements
✅ Supported Models
- All official F5-TTS models, including fine-tuned customs (by suppport them with Custom profile)
- Qwen 2.5/3B Instruct and Microsoft Phi 4 mini instruct for Ai voice chat
📥 Additional Models
- Multilingual (zh & en) (included in project)
- Finnish
- French
- German
- Hindi
- Italian
- Japanese
- Latvian
- Russian
- Russian (included in project)
- Russian (included in project)
- Russian (included in project)
- Russian (included in project)
- Russian (included in project)
- Russian (included in project)
- Spanish
🔥 Very important
- F5-TTS_v1 model is an original F5-TTS model used from own repository (https://huggingface.co/LeeAeron/F5TTSx/models/F5TTS_v1_Base)
- Misha24-10_v2 model is fine-tuned russian model v2 by Misha24-10 used from own repository (https://huggingface.co/LeeAeron/F5TTSx/models/F5TTS_RU/v2)
- Misha24-10_v4 model is fine-tuned russian model v4 (winter) by Misha24-10 used from own repository (https://huggingface.co/LeeAeron/F5TTSx/models/F5TTS_RU/v4)
- ESpeech-TTS-1_podcaster, ESpeech-TTS-1_RL-V1, ESpeech-TTS-1_RL-V2, ESpeech-TTS-1_SFT-95K, ESpeech-TTS-1_SFT-256K are models by ESpeech TTS used from my own repositories.
⚙️ Installation
F5-TTSx uses Python 3.11 and Torch 2.8.0 Cuda 12.8.
F5-TTSx supports GTX and RTX cards, including GTX10xx-16xx and RTX 20xx–50xx.
🖥️ Windows Installation
This project provided with only *.bat installer/re-installer/starter/updater file, that will download and install all components and build fully portable F5-TTSx.
➤ Please Note: - I'm supporting only nVidia GTX10xx-16xx and RTX20xx-50xx GPUs. - This installer is intended for those running Windows 10 or higher. - Application functionality for systems running Windows 7 or lower is not guaranteed.
- Download the F5-TTSx .bat installer for Windows in Releases.
- Place the BAT-file in any folder in the root of any partition with a short Latin name without spaces or special characters and run it.
- Select INSTALL (2) entry .bat file will download, unpack and configure all needed environment.
- After installing, select START (1). .bat will launch Browser, and loads necessary files, models, also there will be loaded Official F5-TTS EN/ZH model.
⚙️ New Features:
- added additional 7 models for usage, deleted E2-TTS model.
- downloadable optional voice pack
- support to change Whisper voice-to-text model, depending your VRAM level
- support for wav/mp3/aac/m4a/m4b/ogg/flac/opus input audio files
- support for change output file format: wav/mp3/aac/m4a/m4b/ogg/flac/opus
- auto-saving synthezed output file into local 'outputs' folder (in project folder)
- support for external .dic dictionaries for right voice pronounce in local 'dicts' folder (in project folder)
- support for external dictionary for right voice pronounce with 'accent_fixes.yaml' file (in project folder)
- support for the pre-placement of accent marks by RUACCENT module for russian language (by Ruaccent button in UI)
- support for the auto-placement of accent marks by RuAccent module for russian language (by RuAccent checkbox in UI, Works for unprepared text in Russian, processing it 'on the fly' during voice synthesis).
- additional Copy/Paste/Clear buttons for 'Text to synteze' (works with clipboard)
- separate additional voice generation settings for the dialogue synthesis mode
📺 Credits
- LeeAeron — additional code, modding, reworking, repository, Hugginface space, features, installer/launcher, reference audios, dictionary support.
- mrfakename — original online demo
- RootingInLoad — chunk generation & podcast app exploration
- jpgallegoar — multiple speech-type generation & voice chat
- Ebany Speech - additional russian language TTS models
📝 License
The F5-TTSx code is released under MIT License.
The pre-trained models are licensed under the CC-BY-NC license due to the training data Emilia, which is an in-the-wild dataset. Sorry for any inconvenience this may cause.