---
configs:
- config_name: train
  data_files:
  - split: train
    path: open_tts_tracker.tsv
  sep: "\t"
- config_name: train_data
  data_files:
  - split: ttsds
    path: data.csv
  sep: ','
license: mit
---
_above models sorted by the amount of capabilities_; [#legend](#legend)

Cloned the GitHub repo for easier viewing and embedding the above table as once requested by @reach-vb: https://github.com/Vaibhavs10/open-tts-tracker/issues/30#issuecomment-1946367525

---

# 🗣️ Open TTS Tracker

A one stop shop to track all open-access/ source Text-To-Speech (TTS) models as they come out. Feel free to make a PR for all those that aren't linked here.

This is aimed as a resource to increase awareness for these models and to make it easier for researchers, developers, and enthusiasts to stay informed about the latest advancements in the field.

> [!NOTE]  
> This repo will only track open source/access codebase TTS models. More motivation for everyone to open-source! 🤗

Some of the models are also being battle tested at TTS arenas:
* 🏆 [TTS Arena](https://huggingface.co/spaces/TTS-AGI/TTS-Arena) - _Battle_ tab allows to choose 2 candidates and compare them
* 🤗🏆 [TTS Spaces Arena](https://huggingface.co/spaces/Pendrokar/TTS-Spaces-Arena) - Uses online HuggingFace Spaces, which have Gradio API enabled

| Name | GitHub | Weights | License | Fine-tune | Languages | Paper | Demo | Issues |
|---|---|---|---|---|---|---|---|---|
| AI4Bharat | [Repo](https://github.com/AI4Bharat/Indic-TTS) | [Hub](https://huggingface.co/ai4bharat) | [MIT](https://github.com/AI4Bharat/Indic-TTS/blob/master/LICENSE.txt) | [Yes](https://github.com/AI4Bharat/Indic-TTS?tab=readme-ov-file#training-steps) | Indic | [Paper](https://arxiv.org/abs/2211.09536) | [Demo](https://models.ai4bharat.org/#/tts) |
| Amphion | [Repo](https://github.com/open-mmlab/Amphion) | [Hub](https://huggingface.co/amphion) | [MIT](https://github.com/open-mmlab/Amphion/blob/main/LICENSE) | No | Multilingual | [Paper](https://arxiv.org/abs/2312.09911) | [🤗 Space](https://huggingface.co/amphion) |  |
| Bark | [Repo](https://github.com/huggingface/transformers/tree/main/src/transformers/models/bark) | [Hub](https://huggingface.co/suno/bark) | [MIT](https://github.com/suno-ai/bark/blob/main/LICENSE) | No | Multilingual | [Paper](https://arxiv.org/abs/2209.03143) | [🤗 Space](https://huggingface.co/spaces/suno/bark) |  |
| EmotiVoice | [Repo](https://github.com/netease-youdao/EmotiVoice) | [GDrive](https://drive.google.com/drive/folders/1y6Xwj_GG9ulsAonca_unSGbJ4lxbNymM) | [Apache 2.0](https://github.com/netease-youdao/EmotiVoice/blob/main/LICENSE) | [Yes](https://github.com/netease-youdao/EmotiVoice/wiki/Voice-Cloning-with-your-personal-data) | ZH + EN | Not Available | Not Available | Separate [GUI agreement](https://github.com/netease-youdao/EmotiVoice/blob/main/EmotiVoice_UserAgreement_%E6%98%93%E9%AD%94%E5%A3%B0%E7%94%A8%E6%88%B7%E5%8D%8F%E8%AE%AE.pdf) |
| F5-TTS | [Repo](https://github.com/SWivid/F5-TTS) | [Hub](https://huggingface.co/SWivid/F5-TTS) | [MIT](https://github.com/SWivid/F5-TTS/blob/master/LICENSE) | Yes | ZH + EN | [Paper](https://arxiv.org/abs/2410.06885) | [🤗 Space](https://huggingface.co/spaces/mrfakename/E2-F5-TTS) |  |
| Fish Speech | [Repo](https://github.com/fishaudio/fish-speech) | [Hub](https://huggingface.co/fishaudio) | [CC-BY-NC-SA 4.0](https://github.com/fishaudio/fish-speech/blob/master/LICENSE) | Yes | Multilingual | Not Available | [🤗 Space](https://huggingface.co/spaces/fishaudio/fish-speech-1) |  |
| Glow-TTS | [Repo](https://github.com/jaywalnut310/glow-tts) | [GDrive](https://drive.google.com/file/d/1JiCMBVTG4BMREK8cT3MYck1MgYvwASL0/view) | [MIT](https://github.com/jaywalnut310/glow-tts/blob/master/LICENSE) | [Yes](https://github.com/jaywalnut310/glow-tts?tab=readme-ov-file#2-pre-requisites) | English | [Paper](https://arxiv.org/abs/2005.11129) | [GH Pages](https://jaywalnut310.github.io/glow-tts-demo/index.html) |  |
| GPT-SoVITS | [Repo](https://github.com/RVC-Boss/GPT-SoVITS) | [Hub](https://huggingface.co/lj1995/GPT-SoVITS) | [MIT](https://github.com/RVC-Boss/GPT-SoVITS/blob/main/LICENSE) | [Yes](https://github.com/RVC-Boss/GPT-SoVITS?tab=readme-ov-file#pretrained-models) | Multilingual | Not Available | Not Available | |
| HierSpeech++ | [Repo](https://github.com/sh-lee-prml/HierSpeechpp) | [GDrive](https://drive.google.com/drive/folders/1-L_90BlCkbPyKWWHTUjt5Fsu3kz0du0w) | [MIT](https://github.com/sh-lee-prml/HierSpeechpp/blob/main/LICENSE) | No | KR + EN | [Paper](https://arxiv.org/abs/2311.12454) | [🤗 Space](https://huggingface.co/spaces/LeeSangHoon/HierSpeech_TTS) | |
| IMS-Toucan | [Repo](https://github.com/DigitalPhonetics/IMS-Toucan) | [GH release](https://github.com/DigitalPhonetics/IMS-Toucan/tags) | [Apache 2.0](https://github.com/DigitalPhonetics/IMS-Toucan/blob/ToucanTTS/LICENSE) | [Yes](https://github.com/DigitalPhonetics/IMS-Toucan#build-a-toucantts-pipeline) | ALL\* | [Paper](https://arxiv.org/abs/2206.12229) | [🤗 Space](https://huggingface.co/spaces/Flux9665/IMS-Toucan), [🤗 Space](https://huggingface.co/spaces/Flux9665/MassivelyMultilingualTTS)\* |  |
| MahaTTS | [Repo](https://github.com/dubverse-ai/MahaTTS) | [Hub](https://huggingface.co/Dubverse/MahaTTS) | [Apache 2.0](https://github.com/dubverse-ai/MahaTTS/blob/main/LICENSE) | No | English + Indic | Not Available | [Recordings](https://github.com/dubverse-ai/MahaTTS/blob/main/README.md#sample-outputs), [Colab](https://colab.research.google.com/drive/1qkZz2km-PX75P0f6mUb2y5e-uzub27NW?usp=sharing) | |
| MaskGCT (Amphion) | [Repo](https://github.com/open-mmlab/Amphion) | [Hub](https://huggingface.co/amphion/MaskGCT) | [CC-BY-NC 4.0](https://huggingface.co/amphion/MaskGCT) | No | Multilingual | [Paper](https://arxiv.org/abs/2409.00750) | [🤗 Space](https://huggingface.co/spaces/amphion/maskgct) |  |
| Matcha-TTS | [Repo](https://github.com/shivammehta25/Matcha-TTS) | [GDrive](https://drive.google.com/drive/folders/17C_gYgEHOxI5ZypcfE_k1piKCtyR0isJ) | [MIT](https://github.com/shivammehta25/Matcha-TTS/blob/main/LICENSE) | [Yes](https://github.com/shivammehta25/Matcha-TTS/tree/main#train-with-your-own-dataset) | English | [Paper](https://arxiv.org/abs/2309.03199) | [🤗 Space](https://huggingface.co/spaces/shivammehta25/Matcha-TTS) | GPL-licensed phonemizer |
| MeloTTS | [Repo](https://github.com/myshell-ai/MeloTTS) | [Hub](https://huggingface.co/myshell-ai) | [MIT](https://github.com/myshell-ai/MeloTTS/blob/main/LICENSE) | Yes | Multilingual | Not Available | [🤗 Space](https://huggingface.co/spaces/mrfakename/MeloTTS) |  |
| MetaVoice-1B | [Repo](https://github.com/metavoiceio/metavoice-src) | [Hub](https://huggingface.co/metavoiceio/metavoice-1B-v0.1/tree/main) | [Apache 2.0](https://github.com/metavoiceio/metavoice-src/blob/main/LICENSE) | [Yes](https://github.com/metavoiceio/metavoice-src?tab=readme-ov-file) | Multilingual | Not Available | [🤗 Space](https://huggingface.co/spaces/mrfakename/MetaVoice-1B-v0.1) |  |
| Neural-HMM TTS | [Repo](https://github.com/shivammehta25/Neural-HMM) | [GitHub](https://github.com/shivammehta25/Neural-HMM/releases) | [MIT](https://github.com/shivammehta25/Neural-HMM/blob/main/LICENSE) | [Yes](https://github.com/shivammehta25/Neural-HMM?tab=readme-ov-file#setup-and-training-using-lj-speech) | English | [Paper](https://arxiv.org/abs/2108.13320) | [GH Pages](https://shivammehta25.github.io/Neural-HMM/) |  |
| OpenVoice | [Repo](https://github.com/myshell-ai/OpenVoice) | [Hub](https://huggingface.co/myshell-ai/OpenVoice) | [MIT](https://github.com/myshell-ai/OpenVoice/blob/main/LICENSE) | No | Multilingual | [Paper](https://arxiv.org/abs/2312.01479) | [🤗 Space](https://huggingface.co/spaces/myshell-ai/OpenVoice) |  |
| OverFlow TTS | [Repo](https://github.com/shivammehta25/OverFlow) | [GitHub](https://github.com/shivammehta25/OverFlow/releases) | [MIT](https://github.com/shivammehta25/OverFlow/blob/main/LICENSE) | [Yes](https://github.com/shivammehta25/OverFlow/tree/main?tab=readme-ov-file#setup-and-training-using-lj-speech) | English | [Paper](https://arxiv.org/abs/2211.06892) | [GH Pages](https://shivammehta25.github.io/OverFlow/) |  |
| Parler TTS | [Repo](https://github.com/huggingface/parler-tts) | [Hub](https://huggingface.co/parler-tts/parler_tts_mini_v0.1) | [Apache 2.0](https://github.com/huggingface/parler-tts/blob/main/LICENSE) | [Yes](https://github.com/huggingface/parler-tts/tree/main/training) | English | Not Available | [🤗 Space](https://huggingface.co/spaces/parler-tts/parler_tts) | |
| pflowTTS | [Unofficial Repo](https://github.com/p0p4k/pflowtts_pytorch) | [GDrive](https://drive.google.com/drive/folders/1x-A2Ezmmiz01YqittO_GLYhngJXazaF0) | [MIT](https://github.com/p0p4k/pflowtts_pytorch/blob/master/LICENSE) | [Yes](https://github.com/p0p4k/pflowtts_pytorch#instructions-to-run) | English | [Paper](https://openreview.net/pdf?id=zNA7u7wtIN) | Not Available | GPL-licensed phonemizer |
| Pheme | [Repo](https://github.com/PolyAI-LDN/pheme) | [Hub](https://huggingface.co/PolyAI/pheme) | [CC-BY](https://github.com/PolyAI-LDN/pheme/blob/main/LICENSE) | [Yes](https://github.com/PolyAI-LDN/pheme#training) | English | [Paper](https://arxiv.org/abs/2401.02839) | [🤗 Space](https://huggingface.co/spaces/PolyAI/pheme) |  |
| Piper | [Repo](https://github.com/rhasspy/piper) | [Hub](https://huggingface.co/datasets/rhasspy/piper-checkpoints/) | [MIT](https://github.com/rhasspy/piper/blob/master/LICENSE.md) | [Yes](https://github.com/rhasspy/piper/blob/master/TRAINING.md) | Multilingual | Not Available | [🤗 Space](https://huggingface.co/spaces/Gregniuki/Pipertts) | [GPL-licensed phonemizer](https://github.com/rhasspy/piper/issues/93) |
| RAD-MMM | [Repo](https://github.com/NVIDIA/RAD-MMM) | [GDrive](https://drive.google.com/file/d/1p8SEVHRlyLQpQnVP2Dc66RlqJVVRDCsJ/view) | [MIT](https://github.com/NVIDIA/RAD-MMM/blob/main/LICENSE) | [Yes](https://github.com/NVIDIA/RAD-MMM?tab=readme-ov-file#training) | Multilingual | [Paper](https://arxiv.org/pdf/2301.10335.pdf) | [Jupyter Notebook](https://github.com/NVIDIA/RAD-MMM/blob/main/inference.ipynb), [Webpage](https://research.nvidia.com/labs/adlr/projects/radmmm/) |  |
| RAD-TTS | [Repo](https://github.com/NVIDIA/radtts) | [GDrive](https://drive.google.com/file/d/1Rb2VMUwQahGrnpFSlAhCPh7OpDN3xgOr/view?usp=sharing) | [MIT](https://github.com/NVIDIA/radtts/blob/main/LICENSE) | [Yes](https://github.com/NVIDIA/radtts#training-radtts-without-pitch-and-energy-conditioning) | English | [Paper](https://openreview.net/pdf?id=0NQwnnwAORi) | [GH Pages](https://nv-adlr.github.io/RADTTS) |  |
| Silero | [Repo](https://github.com/snakers4/silero-models) | [GH links](https://github.com/snakers4/silero-models/blob/master/models.yml) | [CC BY-NC-SA](https://github.com/snakers4/silero-models/blob/master/LICENSE) | [No](https://github.com/snakers4/silero-models/discussions/78) | Multilingual | Not Available | Not Available | [Non Commercial](https://github.com/snakers4/silero-models/wiki/Licensing-and-Tiers) |
| StyleTTS 2 | [Repo](https://github.com/yl4579/StyleTTS2) | [Hub](https://huggingface.co/yl4579/StyleTTS2-LibriTTS/tree/main) | [MIT](https://github.com/yl4579/StyleTTS2/blob/main/LICENSE) | [Yes](https://github.com/yl4579/StyleTTS2#finetuning) | English | [Paper](https://arxiv.org/abs/2306.07691) | [🤗 Space](https://huggingface.co/spaces/styletts2/styletts2) | GPL-licensed phonemizer |
| Tacotron 2 | [Unofficial Repo](https://github.com/NVIDIA/tacotron2) | [GDrive](https://drive.google.com/file/d/1c5ZTuT7J08wLUoVZ2KkUs_VdZuJ86ZqA/view) | [BSD-3](https://github.com/NVIDIA/tacotron2/blob/master/LICENSE) | [Yes](https://github.com/NVIDIA/tacotron2/tree/master?tab=readme-ov-file#training) | English | [Paper](https://arxiv.org/abs/1712.05884) | [Webpage](https://google.github.io/tacotron/publications/tacotron2/) |  |
| TorToiSe TTS | [Repo](https://github.com/neonbjb/tortoise-tts) | [Hub](https://huggingface.co/jbetker/tortoise-tts-v2) | [Apache 2.0](https://github.com/neonbjb/tortoise-tts/blob/main/LICENSE) | [Yes](https://git.ecker.tech/mrq/tortoise-tts) | English | [Technical report](https://arxiv.org/abs/2305.07243) | [🤗 Space](https://huggingface.co/spaces/Manmay/tortoise-tts) |  |
| TTTS | [Repo](https://github.com/adelacvg/ttts) | [Hub](https://huggingface.co/adelacvg/TTTS) | [MPL 2.0](https://github.com/adelacvg/ttts/blob/master/LICENSE) | No | Multilingual | Not Available | [Colab](https://colab.research.google.com/github/adelacvg/ttts/blob/master/demo.ipynb), [🤗 Space](https://huggingface.co/spaces/mrfakename/TTTS) | |
| VALL-E | [Unofficial Repo](https://github.com/enhuiz/vall-e) | Not Available | [MIT](https://github.com/enhuiz/vall-e/blob/main/LICENSE) | [Yes](https://github.com/enhuiz/vall-e#get-started) | NA | [Paper](https://arxiv.org/abs/2301.02111) | Not Available |  |
| VITS/ MMS-TTS | [Repo](https://github.com/huggingface/transformers/tree/7142bdfa90a3526cfbed7483ede3afbef7b63939/src/transformers/models/vits) | [Hub](https://huggingface.co/kakao-enterprise) / [MMS](https://huggingface.co/models?search=mms-tts) | [Apache 2.0](https://github.com/huggingface/transformers/blob/main/LICENSE) | [Yes](https://github.com/ylacombe/finetune-hf-vits) | English | [Paper](https://arxiv.org/abs/2106.06103) | [🤗 Space](https://huggingface.co/spaces/kakao-enterprise/vits) | GPL-licensed phonemizer |
| WhisperSpeech | [Repo](https://github.com/collabora/WhisperSpeech) | [Hub](https://huggingface.co/collabora/whisperspeech) | [MIT](https://github.com/collabora/WhisperSpeech/blob/main/LICENSE) | No | Multilingual | Not Available | [🤗 Space](https://huggingface.co/spaces/collabora/WhisperSpeech), [Recordings](https://github.com/collabora/WhisperSpeech/blob/main/README.md), [Colab](https://colab.research.google.com/github/collabora/WhisperSpeech/blob/8168a30f26627fcd15076d10c85d9e33c52204cf/Inference%20example.ipynb) | |
| XTTS | [Repo](https://github.com/coqui-ai/TTS) | [Hub](https://huggingface.co/coqui/XTTS-v2) | [CPML](https://coqui.ai/cpml) | [Yes](https://docs.coqui.ai/en/latest/models/xtts.html#training) | Multilingual | [Paper](https://arxiv.org/abs/2406.04904) | [🤗 Space](https://huggingface.co/spaces/coqui/xtts) | Non Commercial |
| xVASynth | [Repo](https://github.com/DanRuta/xVA-Synth) | [Hub](https://huggingface.co/Pendrokar/xvapitch) | [GPL-3.0](https://github.com/DanRuta/xVA-Synth/blob/master/LICENSE.md) | [Yes](https://github.com/DanRuta/xva-trainer) | Multilingual | Not Available | [🤗 Space](https://huggingface.co/spaces/Pendrokar/xVASynth) | Base model trained on non-permissive datasets |

* *Multilingual* - Amount of supported languages is ever changing, check the Space and Hub which specific languages are supported
* *ALL* - Supports all natural languages; may not support artificial/contructed languages

Also to find a model for a specific language, filter out the TTS models hosted on HuggingFace: <https://huggingface.co/models?pipeline_tag=text-to-speech&language=en&sort=trending>

---

## Legend

For the [#above](#) TTS capability table. Open the [viewer](../../viewer/) in another window or even another monitor to keep both it and the legend in view.

* Processor ⚡ - Inference done by
  * CPU (CPU**s** = multithreaded) - All models can be run on CPU, so real-time factor should be below 2.0 to qualify for CPU tag, though some more leeway can be given if it supports audio streaming
  * CUDA by *NVIDIA*™
  * ROCm by *AMD*™, also see [ONNX Runtime HF guide](https://huggingface.co/docs/optimum/en/onnxruntime/usage_guides/amdgpu)
* Phonetic alphabet 🔤 - Phonetic transcription that allows to control pronunciation of words before inference
  * [IPA](https://en.wikipedia.org/wiki/International_Phonetic_Alphabet) - International Phonetic Alphabet
  * [ARPAbet](https://en.wikipedia.org/wiki/ARPABET) - American English focused phonetics
* Insta-clone 👥 - Zero-shot model for quick voice cloning
* Emotion control 🎭 - Able to force an emotional state of speaker
  * 🎭 <# emotions> ( 😡 anger; 😃 happiness; 😭 sadness; 😯 surprise; 🤫 whispering; 😊 friendlyness )
  * 🎭👥 strict insta-clone switch - cloned on sample with specific emotion; may sound different than normal speaking voice; no ability to go in-between states
  * 🎭📖 strict control through prompt - prompt input parameter
* Prompting 📖 - Also a side effect of narrator based datasets and a way to affect the emotional state
  * 📖 - Prompt as a separate input parameter
  * 🗣📖 - The prompt itself is also spoken by TTS; [ElevenLabs docs](https://elevenlabs.io/docs/speech-synthesis/prompting#emotion)
* Streaming support 🌊 - Can playback audio while it is still being generated
* Speech control 🎚 - Ability to change the pitch, duration, etc. for the whole and/or per-phoneme of the generated speech
* Voice conversion / Speech-To-Speech 🦜 - Streaming support implies real-time S2S; S2T=>T2S does not count
* Longform synthesis 📜 - Able to synthesize whole paragraphs, as some TTS models tend to break down after a certain audio length limit

Example if the proprietary ElevenLabs were to be added to the capabilities table:
| Name | Processor<br>⚡ | Phonetic alphabet<br>🔤 | Insta-clone<br>👥 | Emotional control<br>🎭 | Prompting<br>📖 | Speech control<br>🎚 | Streaming support<br>🌊 | Voice conversion<br>🦜 | Longform synthesis<br>📜 |
|---|---|---|---|---|---|---|---|---| --- |
|ElevenLabs|CUDA|IPA, ARPAbet|👥|🎭📖|🗣📖|🎚 stability, voice similarity|🌊|🦜|📜 Projects|

More info on how the capabilities table came about can be found within the [GitHub Issue](https://github.com/Vaibhavs10/open-tts-tracker/issues/14).

## train_data Legend

Legend for the separate TTSDS Datasets ([_train_data_ viewer](https://huggingface.co/datasets/Pendrokar/open_tts_tracker/viewer/train_data) [GitHub](https://github.com/ttsds/ttsds_systems))

- 🌐 Multilingual
  - The ISO codes of languages the model is capable off. ❌ if English only.
- 📚 Training Amount (k hours)
  - The number of hours the model was trained on
- 🧠 Num. Parameters (M)
  - How many parameters the model has, excluding vocoder and text-only components
- 🎯 Target Repr.
  - Which output representations the model uses, for example audio codecs or mel spectrograms
- 📖 LibriVox Only
  - If the model was trained on librivox-like (audiobook) data alone
- 🔄 NAR
  - If the model has a significant non-autoregressive component
- 🔁 AR
  - If the model has a significant autoregressive component
- 🔡 G2P
  - If the model uses G2P (phone inputs)
- 🧩 Language Model
  - If an LM-like approach is used (next token prediction)
- 🎵 Prosody Prediction
  - If prosodic correlates such as pitch or energy are predicted
- 🌊 Diffusion
  - If diffusion is used (outside vocoder)
- ⏱️ Delay Pattern
  - If a delay pattern is used for audio codes (see [Lyth & King, 2024](https://arxiv.org/abs/2402.01912))

_Please create pull requests to update the info on the models!_

---