DDSP-SVC

Language: English 简体中文 한국어

DDSP-SVC

<div align="center"> <img src="https://storage.googleapis.com/ddsp/github_images/ddsp_logo.png" width="200px" alt="logo"></img> </div> Real-time end-to-end singing voice conversion system based on DDSP (Differentiable Digital Signal Processing）.

(3.0 - Experimental) Shallow diffusion model (DDSP + Diff-SVC refactor version)

Data preparation, configuring the pre-trained encoder (hubert or contentvec ) and vocoder (nsf-hifigan) is the same as training a pure DDSP model.

Because the diffusion model is more difficult to train, we provide some pre-trained models here:

https://huggingface.co/datasets/ms903/Diff-SVC-refactor-pre-trained-model/blob/main/hubertsoft_pitch_410k/model_0.pt (using 'hubertsoft' encoder)

https://huggingface.co/datasets/ms903/Diff-SVC-refactor-pre-trained-model/blob/main/pitch_400k/model_0.pt (using 'contentvec768l12' encoder)

Move the model_0.pt to the model export folder specified by the 'expdir' parameter in diffusion.yaml, and the program will automatically load the pre-trained models in that folder.

(1) Preprocessing：

python preprocess.py -c configs/diffusion.yaml

This preprocessing can also be used to train the DDSP model without preprocessing twice, but you need to ensure that the parameters under the 'data' tag in yaml files are consistent.

(2) Train a diffusion model：

python train_diff.py -c configs/diffusion.yaml

(3) Train a DDSP model：

python train.py -c configs/combsub.yaml

As mentioned above, re-preprocessing is not required, but please check whether the parameters of combsub.yaml and diffusion.yaml match. The number of speakers 'n_spk' can be inconsistent, but try to use the same number to represent the same speaker (this makes inference easier).

(4) Non-real-time inference：

python main_diff.py -i <input.wav> -ddsp <ddsp_ckpt.pt> -diff <diff_ckpt.pt> -o <output.wav> -k <keychange (semitones)> -id <speaker_id> -diffid <diffusion_speaker_id> -speedup <speedup> -method <method> -kstep <kstep>

'speedup' is the acceleration speed, 'method' is 'pndm' or 'dpm-solver', 'kstep' is the number of shallow diffusion steps, 'diffid' is the speaker id of the diffusion model, and other parameters have the same meaning as main.py.

If the same id has been used to represent the same speaker during training, '-diffid' can be empty, otherwise the '-diffid' option needs to be specified.

If '-ddsp' is empty, the pure diffusion model is used, at this time, shallow diffusion is performed with the mel of the input source, and if further '-kstep' is empty, full-depth Gaussian diffusion is performed.

The program will automatically check whether the parameters of the DDSP model and the diffusion model match (sampling rate, hop size and encoder), and if they do not match, it will ignore loading the DDSP model and enter Gaussian diffusion mode.

Real-time GUI：

python gui_diff.py

0. Introduction

DDSP-SVC is a new open source singing voice conversion project dedicated to the development of free AI voice changer software that can be popularized on personal computers.

Compared with the more famous Diff-SVC and SO-VITS-SVC, its training and synthesis have much lower requirements for computer hardware, and the training time can be shortened by orders of magnitude. In addition, when performing voice change in real-time, the hardware resources of this project are significantly lower than SO-VITS-SVC, and Diff-SVC is too slow to perform voice change in real-time.

Although the original synthesis quality of DDSP is not ideal (the original output can be heard in tensorboard while training), after using the pre-trained vocoder-based enhancer, the sound quality for some dateset can reach a level close to SO-VITS-SVC.

If the quality of the training data is very high, probably still Diff-SVC will have the highest sound quality. The demo outputs are in the samples folder, and the related model checkpoint can be downloaded from the release page.

Disclaimer: Please make sure to only train DDSP-SVC models with legally obtained authorized data, and do not use these models and any audio they synthesize for illegal purposes. The author of this repository is not responsible for any infringement, fraud and other illegal acts caused by the use of these model checkpoints and audio.

Update log: I am too lazy to translate, please see the Chinese version readme.

1. Installing the dependencies

We recommend first installing PyTorch from the official website, then run:

pip install -r requirements.txt 

NOTE : I only test the code using python 3.8 (windows) + torch 1.9.1 + torchaudio 0.6.0, too new or too old dependencies may not work

UPDATE: python 3.8 (windows) + cuda 11.8 + torch 2.0.0 + torchaudio 2.0.1 works, and training is faster.

2. Configuring the pretrained model

UPDATE: ContentVec encoder is supported now. You can download the pretrained ContentVec encoder instead of HubertSoft encoder and modify the configuration file to use it.

• (Required) Download the pretrained HubertSoft encoder and put it under pretrain/hubert folder.
• Get the pretrained vocoder-based enhancer from the DiffSinger Community Vocoders Project and unzip it into pretrain/ folder

3. Preprocessing

Put all the training dataset (.wav format audio clips) in the below directory: data/train/audio. Put all the validation dataset (.wav format audio clips) in the below directory: data/val/audio. You can also run

python draw.py

to help you select validation data (you can adjust the parameters in draw.py to modify the number of extracted files and other parameters)

Then run

python preprocess.py -c configs/combsub.yaml

for a model of combtooth substractive synthesiser (recommend), or run

python preprocess.py -c configs/sins.yaml

for a model of sinusoids additive synthesiser.

You can modify the configuration file config/<model_name>.yaml before preprocessing. The default configuration is suitable for training 44.1khz high sampling rate synthesiser with GTX-1660 graphics card.

NOTE 1: Please keep the sampling rate of all audio clips consistent with the sampling rate in the yaml configuration file ! If it is not consistent, the program can be executed safely, but the resampling during the training process will be very slow.

NOTE 2: The total number of the audio clips for training dataset is recommended to be about 1000, especially long audio clip can be cut into short segments, which will speed up the training, but the duration of all audio clips should not be less than 2 seconds. If there are too many audio clips, you need a large internal-memory or set the 'cache_all_data' option to false in the configuration file.

NOTE 3: The total number of the audio clips for validation dataset is recommended to be about 10, please don't put too many or it will be very slow to do the validation.

NOTE 4: If your dataset is not very high quality, set 'f0_extractor' to 'crepe' in the config file. The crepe algorithm has the best noise immunity, but at the cost of greatly increasing the time required for data preprocessing.

UPDATE: Multi-speaker training is supported now. The 'n_spk' parameter in configuration file controls whether it is a multi-speaker model. If you want to train a multi-speaker model, audio folders need to be named with positive integers not greater than 'n_spk' to represent speaker ids, the directory structure is like below:

# training dataset
# the 1st speaker
data/train/audio/1/aaa.wav
data/train/audio/1/bbb.wav
...
# the 2nd speaker
data/train/audio/2/ccc.wav
data/train/audio/2/ddd.wav
...

# validation dataset
# the 1st speaker
data/val/audio/1/eee.wav
data/val/audio/1/fff.wav
...
# the 2nd speaker
data/val/audio/2/ggg.wav
data/val/audio/2/hhh.wav
...

If 'n_spk' = 1, The directory structure of the single speaker model is still supported, which is like below:

# training dataset
data/train/audio/aaa.wav
data/train/audio/bbb.wav
...
# validation dataset
data/val/audio/ccc.wav
data/val/audio/ddd.wav
...

4. Training

# train a combsub model as an example
python train.py -c configs/combsub.yaml

The command line for training other models is similar.

You can safely interrupt training, then running the same command line will resume training.

You can also finetune the model if you interrupt training first, then re-preprocess the new dataset or change the training parameters (batchsize, lr etc.) and then run the same command line.

5. Visualization

# check the training status using tensorboard
tensorboard --logdir=exp

Test audio samples will be visible in TensorBoard after the first validation.

NOTE: The test audio samples in Tensorboard are the original outputs of your DDSP-SVC model that is not enhanced by an enhancer. If you want to test the synthetic effect after using the enhancer (which may have higher quality) , please use the method described in the following chapter.

6. Non-real-time VC

(Recommend) Enhance the output using the pretrained vocoder-based enhancer:

# high audio quality in the normal vocal range if enhancer_adaptive_key = 0 (default)
# set enhancer_adaptive_key > 0 to adapt the enhancer to a higher vocal range
python main.py -i <input.wav> -m <model_file.pt> -o <output.wav> -k <keychange (semitones)> -id <speaker_id> -eak <enhancer_adaptive_key (semitones)>

Raw output of DDSP:

# fast, but relatively low audio quality (like you hear in tensorboard)
python main.py -i <input.wav> -m <model_file.pt> -o <output.wav> -k <keychange (semitones)> -id <speaker_id> -e false

Other options about the f0 extractor and response threhold，see:

python main.py -h

(UPDATE) Mix-speaker is supported now. You can use "-mix" option to design your own vocal timbre, below is an example:

# Mix the timbre of 1st and 2nd speaker in a 0.5 to 0.5 ratio
python main.py -i <input.wav> -m <model_file.pt> -o <output.wav> -k <keychange (semitones)> -mix "{1:0.5, 2:0.5}" -eak 0

7. Real-time VC

Start a simple GUI with the following command:

python gui.py

The front-end uses technologies such as sliding window, cross-fading, SOLA-based splicing and contextual semantic reference, which can achieve sound quality close to non-real-time synthesis with low latency and resource occupation.

Update: A splicing algorithm based on a phase vocoder is now added, but in most cases the SOLA algorithm already has high enough splicing sound quality, so it is turned off by default. If you are pursuing extreme low-latency real-time sound quality, you can consider turning it on and tuning the parameters carefully, and there is a possibility that the sound quality will be higher. However, a large number of tests have found that if the cross-fade time is longer than 0.1 seconds, the phase vocoder will cause a significant degradation in sound quality.

8. Acknowledgement

• ddsp
• pc-ddsp
• soft-vc
• DiffSinger (OpenVPI version)