This fork adds in:
- v1 seed-uvit-whisper-small-wavenet and seed-uvit-whisper-base to the Realtime GUI
I haven't completely fleshed out the implementation of it, there still seem to be some bugs on inference maybe repeating and even though there is an option for v2 models, it doesn't 8000 work. All of the modifications were done with cursor utilizing o4-mini, basing the inference off of the respective app.py files that already work for these models.
real-time-demo.webm
Currently released model supports zero-shot voice conversion 🔊 , zero-shot real-time voice conversion 🗣️ and zero-shot singing voice conversion 🎶. Without any training, it is able to clone a voice given a reference speech of 1~30 seconds.
We support further fine-tuning on custom data to increase performance on specific speaker/speakers, with extremely low data requirement (minimum 1 utterance per speaker) and extremely fast training speed (minimum 100 steps, 2 min on T4)!
Real-time voice conversion is support, with algorithm delay of ~300ms and device side delay of ~100ms, suitable for online meetings, gaming and live streaming.
To find a list of demos and comparisons with previous voice conversion models, please visit our demo page🌐 and Evaluaiton📊.
We are keeping on improving the model quality and adding more features.
See EVAL.md for objective evaluation results and comparisons with other baselines.
Suggested python 3.10 on Windows, Mac M Series (Apple Silicon) or Linux. Windows and Linux:
pip install -r requirements.txtMac M Series:
pip install -r requirements-mac.txtFor Windows users, you may consider install triton-windows to enable --compile usage, which gains speed up on V2 models:
pip install triton-windows==3.2.0.post13We have released 4 models for different purposes:
| Version | Name | Purpose | Sampling Rate | Content Encoder | Vocoder | Hidden Dim | N Layers | Params | Remarks |
|---|---|---|---|---|---|---|---|---|---|
| v1.0 | seed-uvit-tat-xlsr-tiny (🤗📄) | Voice Conversion (VC) | 22050 | XLSR-large | HIFT | 384 | 9 | 25M | suitable for real-time voice conversion |
| v1.0 | seed-uvit-whisper-small-wavenet (🤗📄) | Voice Conversion (VC) | 22050 | Whisper-small | BigVGAN | 512 | 13 | 98M | suitable for offline voice conversion |
| v1.0 | seed-uvit-whisper-base (🤗📄) | Singing Voice Conversion (SVC) | 44100 | Whisper-small | BigVGAN | 768 | 17 | 200M | strong zero-shot performance, singing voice conversion |
| v2.0 | hubert-bsqvae-small (🤗📄) | Voice & Accent Conversion (VC) | 22050 | ASTRAL-Quantization | BigVGAN | 512 | 13 | 67M(CFM) + 90M(AR) | Best in suppressing source speaker traits |
Checkpoints of the latest model release will be downloaded automatically when first run inference.
If you are unable to access huggingface for network reason, try using mirror by adding HF_ENDPOINT=https://hf-mirror.com before every command.
Command line inference:
python inference.py --source <source-wav>
--target <referene-wav>
--output <output-dir>
--diffusion-steps 25 # recommended 30~50 for singingvoice conversion
--length-adjust 1.0
--inference-cfg-rate 0.7
--f0-condition False # set to True for singing voice conversion
--auto-f0-adjust False # set to True to auto adjust source pitch to target pitch level, normally not used in singing voice conversion
--semi-tone-shift 0 # pitch shift in semitones for singing voice conversion
--checkpoint <path-to-checkpoint>
--config <path-to-config>
--fp16 Truewhere:
sourceis the path to the speech file to convert to reference voicetargetis the path to the speech file as voice referenceoutputis the path to the output directorydiffusion-stepsis the number of diffusion steps to use, default is 25, use 30-50 for best quality, use 4-10 for fastest inferencelength-adjustis the length adjustment factor, default is 1.0, set <1.0 for speed-up speech, >1.0 for slow-down speechinference-cfg-ratehas subtle difference in the output, default is 0.7f0-conditionis the flag to condition the pitch of the output to the pitch of the source audio, default is False, set to True for singing voice conversionauto-f0-adjustis the flag to auto adjust source pitch to target pitch level, default is False, normally not used in singing voice conversionsemi-tone-shiftis the pitch shift in semitones for singing voice conversion, default is 0checkpointis the path to the model checkpoint if you have trained or fine-tuned your own model, leave to blank to auto-download default model from huggingface.(seed-uvit-whisper-small-wavenetiff0-conditionisFalseelseseed-uvit-whisper-base)configis the path to the model config if you have trained or fine-tuned your own model, leave to blank to auto-download default config from huggingfacefp16is the flag to use float16 inference, default is True
Similarly, to use V2 model, you can run:
python inference_v2.py --source <source-wav>
--target <referene-wav>
--output <output-dir>
--diffusion-steps 25 # recommended 30~50 for singingvoice conversion
--length-adjust 1.0 # same as V1
--intelligibility-cfg-rate 0.7 # controls how clear the output linguistic content is, recommended 0.0~1.0
--similarity-cfg-rate 0.7 # controls how similar the output voice is to the reference voice, recommended 0.0~1.0
--convert-style true # whether to use AR model for accent & emotion conversion, set to false will only conduct timbre conversion similar to V1
--anonymization-only false # set to true will ignore reference audio but only anonymize source speech to an "average voice"
--top-p 0.9 # controls the diversity of the AR model output, recommended 0.5~1.0
--temperature 1.0 # controls the randomness of the AR model output, recommended 0.7~1.2
--repetition-penalty 1.0 # penalizes the repetition of the AR model output, recommended 1.0~1.5
--cfm-checkpoint-path <path-to-cfm-checkpoint> # path to the checkpoint of the CFM model, leave to blank to auto-download default model from huggingface
--ar-checkpoint-path <path-to-ar-checkpoint> # path to the checkpoint of the AR model, leave to blank to auto-download default model from huggingfaceVoice Conversion Web UI:
python app_vc.py --checkpoint <path-to-checkpoint> --config <path-to-config> --fp16 Truecheckpointis the path to the model checkpoint if you have trained or fine-tuned your own model, leave to blank to auto-download default model from huggingface. (seed-uvit-whisper-small-wavenet)configis the path to the model config if you have trained or fine-tuned your own model, leave to blank to auto-download default config from huggingface
Then open the browser and go to http://localhost:7860/ to use the web interface.
Singing Voice Conversion Web UI:
python app_svc.py --checkpoint <path-to-checkpoint> --config <path-to-config> --fp16 Truecheckpointis the path to the model checkpoint if you have trained or fine-tuned your own model, leave to blank to auto-download default model from huggingface. (seed-uvit-whisper-base)configis the path to the model config if you have trained or fine-tuned your own model, leave to blank to auto-download default config from huggingface
V2 model Web UI:
python app_vc_v2.py --cfm-checkpoint-path <path-to-cfm-checkpoint> --ar-checkpoint-path <path-to-ar-checkpoint>cfm-checkpoint-pathis the path to the checkpoint of the CFM model, leave to blank to auto-download default model from huggingfacear-checkpoint-pathis the path to the checkpoint of the AR model, leave to blank to auto-download default model from huggingface- you may consider adding
--compileto gain ~x6 speed-up on AR model inference
Integrated Web UI:
python app.py --enable-v1 --enable-v2This will only load pretrained models for zero-shot inference. To use custom checkpoints, please run app_vc.py or app_svc.py as above.
If you have limited memory, remove --enable-v2 or --enable-v1 to only load one of the model sets.
Real-time voice conversion GUI:
python real-time-gui.py --checkpoint-path <path-to-checkpoint> --config-path <path-to-config>checkpointis the path to the model checkpoint if you have trained or fine-tuned your own model, leave to blank to auto-download default model from huggingface. (seed-uvit-tat-xlsr-tiny)configis the path to the model config if you have trained or fine-tuned your own model, leave to blank to auto-download default config from huggingface
Important
It is strongly recommended to use a GPU for real-time voice conversion. Some performance testing has been done on a NVIDIA RTX 3060 Laptop GPU, results and recommended parameter settings are listed below:
| Model Configuration | Diffusion Steps | Inference CFG Rate | Max Prompt Length | Block Time (s) | Crossfade Length (s) | Extra context (left) (s) | Extra context (right) (s) | Latency (ms) | Inference Time per Chunk (ms) |
|---|---|---|---|---|---|---|---|---|---|
| seed-uvit-xlsr-tiny | 10 | 0.7 | 3.0 | 0.18s | 0.04s | 2.5s | 0.02s | 430ms | 150ms |
You can adjust the parameters in the GUI according to your own device performance, the voice conversion stream should work well as long as Inference Time is less than Block Time.
Note that inference speed may drop if you are running other GPU intensive tasks (e.g. gaming, watching videos)
Explanations for real-time voice conversion GUI parameters:
Diffusion Stepsis the number of diffusion steps to use, in real-time case usually set to 4~10 for fastest inference;Inference CFG Ratehas subtle difference in the output, default is 0.7, set to 0.0 gains about 1.5x speed-up;Max Prompt Lengthis the maximum length of the prompt audio, setting to a low value can speed up inference, but may reduce similarity to prompt speech;Block Timeis the time length of each audio chunk for inference, the higher the value, the higher the latency, note this value must be greater than the inference time per block, set according to your hardware condition;Crossfade Lengthis the time length of crossfade between audio chunks, normally not needed to change;Extra context (left)is the time length of extra history context for inference, the higher the value, the higher the inference time, but can increase stability;Extra context (right)is the time length of extra future context for inference, the higher the value, the higher the inference time and latency, but can increase stability;
The algorithm delay is appoximately calculated as Block Time * 2 + Extra context (right), device side delay is usually of ~100ms. The overall delay is the sum of the two.
You may wish to use VB-CABLE to route audio from GUI output stream to a virtual microphone.
(GUI and audio chunking logic are modified from RVC, thanks for their brilliant implementation!)
Fine-tuning on custom data allow the model to clone someone's voice more accurately. It will largely improve speaker similarity on particular speakers, but may slightly increase WER.
A Colab Tutorial is here for you to follow:
- Prepare your own dataset. It has to satisfy the following:
- File structure does not matter
- Each audio file should range from 1 to 30 seconds, otherwise will be ignored
- All audio files should be in on of the following formats:
.wav.flac.mp3.m4a.opus.ogg - Speaker label is not required, but make sure that each speaker has at least 1 utterance
- Of course, the more data you have, the better the model will perform
- Training data should be as clean as possible, BGM or noise is not desired
- Choose a model configuration file from
configs/presets/for fine-tuning, or create your own to train from scratch.- For fine-tuning, it should be one of the following:
./configs/presets/config_dit_mel_seed_uvit_xlsr_tiny.ymlfor real-time voice conversion./configs/presets/config_dit_mel_seed_uvit_whisper_small_wavenet.ymlfor offline voice conversion./configs/presets/config_dit_mel_seed_uvit_whisper_base_f0_44k.ymlfor singing voice conversion
- For fine-tuning, it should be one of the following:
- Run the following command to start training:
python train.py
--config <path-to-config>
--dataset-dir <path-to-data>
--run-name <run-name>
--batch-size 2
--max-steps 1000
--max-epochs 1000
--save-every 500
--num-workers 0where:
configis the path to the model config, choose one of the above for fine-tuning or create your own for training from scratchdataset-diris the path to the dataset directory, which should be a folder containing all the audio filesrun-nameis the name of the run, which will be used to save the model checkpoints and logsbatch-sizeis the batch size for training, choose depends on your GPU memory.max-stepsis the maximum number of steps to train, choose depends on your dataset size and training timemax-epochsis the maximum number of epochs to train, choose depends on your dataset size and training timesave-everyis the number of steps to save the model checkpointnum-workersis the number of workers for data loading, set to 0 for Windows
Similarly, to train V2 model, you can run: (note that V2 training script supports multi-GPU training)
accelerate launch train_v2.py
--dataset-dir <path-to-data>
--run-name <run-name>
--batch-size 2
--max-steps 1000
871A
--max-epochs 1000
--save-every 500
--num-workers 0
--train-cfm-
If training accidentially stops, you can resume training by running the same command again, the training will continue from the last checkpoint. (Make sure
run-nameandconfigarguments are the same so that latest checkpoint can be found) -
After training, you can use the trained model for inference by specifying the path to the checkpoint and config file.
- They should be under
./runs/<run-name>/, with the checkpoint namedft_model.pthand config file with the same name as the training config file. - You still have to specify a reference audio file of the speaker you'd like to use during inference, similar to zero-shot usage.
- They should be under
- Release code
- Release pretrained models:
- Huggingface space demo:
- HTML demo page: Demo
- Streaming inference
- Reduce streaming inference latency
- Demo video for real-time voice conversion
- Singing voice conversion
- Noise resiliency for source audio
- Potential architecture improvements
- U-ViT style skip connections
- Changed input to OpenAI Whisper
- Time as Token
- Code for training on custom data
- Few-shot/One-shot speaker fine-tuning
- Changed to BigVGAN from NVIDIA for singing voice decoding
- Whisper version model for singing voice conversion
- Objective evaluation and comparison with RVC/SoVITS for singing voice conversion
- Improve audio quality
- NSF vocoder for better singing voice conversion
- Fix real-time voice conversion artifact while not talking (done by adding a VAD model)
- Colab Notebook for fine-tuning example
- Replace whisper with more advanced linguistic content extractor
- More to be added
- Add Apple Silicon support
- Release paper, evaluations and demo page for V2 model
- On Mac - running
real-time-gui.pymight raise an errorModuleNotFoundError: No module named '_tkinter', in this case a new Python version with Tkinter support should be installed. Refer to This Guide on stack overflow for explanation of the problem and a detailed fix.
- 2024-04-16
- Released V2 model for voice and accent conversion, with better anonymization of source speaker
- 2025-03-03:
- Added Mac M Series (Apple Silicon) support
- 2024-11-26:
- Updated v1.0 tiny version pretrained model, optimized for real-time voice conversion
- Support one-shot/few-shot single/multi speaker fine-tuning
- Support using custom checkpoint for webUI & real-time GUI
- 2024-11-19:
- arXiv paper released
- 2024-10-28:
- Updated fine-tuned 44k singing voice conversion model with better audio quality
- 2024-10-27:
- Added real-time voice conversion GUI
- 2024-10-25:
- Added exhaustive evaluation results and comparisons with RVCv2 for singing voice conversion
- 2024-10-24:
- Updated 44kHz singing voice conversion model, with OpenAI Whisper as speech content input
- 2024-10-07:
- Updated v0.3 pretrained model, changed speech content encoder to OpenAI Whisper
- Added objective evaluation results for v0.3 pretrained model
- 2024-09-22:
- Updated singing voice conversion model to use BigVGAN from NVIDIA, providing large improvement to high-pitched singing voices
- Support chunking and streaming output for long audio files in Web UI
- 2024-09-18:
- Updated f0 conditioned model for singing voice conversion
- 2024-09-14:
- Updated v0.2 pretrained model, with smaller size and less diffusion steps to achieve same quality, and additional ability to control prosody preservation
- Added command line inference script
- Added installation and usage instructions
- Amphion for providing computational resources and inspiration!
- Vevo for theoretical foundation of V2 model
- MegaTTS3 for multi-condition CFG inference implemented in V2 model
- ASTRAL-quantiztion for the amazing speaker-disentangled speech tokenizer used by V2 model
- RVC for foundationing the real-time voice conversion
- SEED-TTS for the initial idea