@@ -642,8 +642,7 @@ Audio samples generated by a TTS system. Text is first transformed into spectrog
Multi-Speaker TTS
-------------------
PaddleSpeech also support Multi-Speaker TTS, we provide the audio demos generated by FastSpeech2 + ParallelWaveGAN, we use AISHELL-3 Multi-Speaker TTS dataset.
PaddleSpeech also support Multi-Speaker TTS, we provide the audio demos generated by FastSpeech2 + ParallelWaveGAN, we use AISHELL-3 Multi-Speaker TTS dataset. Each line is a different person.
.. raw:: html
...
...
@@ -651,10 +650,370 @@ PaddleSpeech also support Multi-Speaker TTS, we provide the audio demos generate
You can find PaddleSpeech TTS's tacotron2 with LJSpeech dataset example at [examples/ljspeech/tts0](https://github.com/PaddlePaddle/DeepSpeech/tree/develop/examples/ljspeech/tts0).
You can find PaddleSpeech TTS's tacotron2 with LJSpeech dataset example at [examples/ljspeech/tts0](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/ljspeech/tts0).
### TransformerTTS
**Disadvantages of the Tacotrons:**
...
...
@@ -118,7 +118,7 @@ Transformer TTS is a combination of Tacotron2 and Transformer.
@@ -146,7 +146,7 @@ Transformer TTS is a seq2seq acoustic model based on Transformer and Tacotron2.
- The ability to perceive local information is weak, and local information is more related to pronunciation.
- Stability is worse than Tacotron2.
You can find PaddleSpeech TTS's Transformer TTS with LJSpeech dataset example at [examples/ljspeech/tts1](https://github.com/PaddlePaddle/DeepSpeech/tree/develop/examples/ljspeech/tts1).
You can find PaddleSpeech TTS's Transformer TTS with LJSpeech dataset example at [examples/ljspeech/tts1](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/ljspeech/tts1).
### FastSpeech2
...
...
@@ -184,14 +184,14 @@ Instead of using the encoder-attention-decoder based architecture as adopted by
• Can be generated in parallel (decoding time is less affected by sequence length)
[FastPitch](https://arxiv.org/abs/2006.06873) follows FastSpeech. A single pitch value is predicted for every temporal location, which improves the overall quality of synthesized speech.
You can find PaddleSpeech TTS's FastSpeech2/FastPitch with CSMSC dataset example at [examples/csmsc/tts3](https://github.com/PaddlePaddle/DeepSpeech/tree/develop/examples/csmsc/tts3), We use token-averaged pitch and energy values introduced in FastPitch rather than frame level ones in FastSpeech2.
You can find PaddleSpeech TTS's FastSpeech2/FastPitch with CSMSC dataset example at [examples/csmsc/tts3](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/csmsc/tts3), We use token-averaged pitch and energy values introduced in FastPitch rather than frame level ones in FastSpeech2.
### SpeedySpeech
[SpeedySpeech](https://arxiv.org/abs/2008.03802) simplify the teacher-student architecture of FastSpeech and provide a fast and stable training procedure.
...
...
@@ -223,10 +223,10 @@ You can find PaddleSpeech TTS's FastSpeech2/FastPitch with CSMSC dataset example
- Describe a simple data augmentation technique that can be used early in the training to make the teacher network robust to sequential error propagation.
You can find PaddleSpeech TTS's SpeedySpeech with CSMSC dataset example at [examples/csmsc/tts2](https://github.com/PaddlePaddle/DeepSpeech/tree/develop/examples/csmsc/tts2).
You can find PaddleSpeech TTS's SpeedySpeech with CSMSC dataset example at [examples/csmsc/tts2](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/csmsc/tts2).
## Vocoders
In speech synthesis, the main task of the vocoder is to convert the spectral parameters predicted by the acoustic model into the final speech waveform.
...
...
@@ -276,7 +276,7 @@ Here, we introduce a Flow-based vocoder WaveFlow and a GAN-based vocoder Paralle
- It is a small-footprint flow-based model for raw audio. It has only 5.9M parameters, which is 15x smalller than WaveGlow (87.9M).
- It is directly trained with maximum likelihood without probability density distillation and auxiliary losses as used in [Parallel WaveNet](https://arxiv.org/abs/1711.10433) and [ClariNet](https://openreview.net/pdf?id=HklY120cYm), which simplifies the training pipeline and reduces the cost of development.
You can find PaddleSpeech TTS's WaveFlow with LJSpeech dataset example at [examples/ljspeech/voc0](https://github.com/PaddlePaddle/DeepSpeech/tree/develop/examples/ljspeech/voc0).
You can find PaddleSpeech TTS's WaveFlow with LJSpeech dataset example at [examples/ljspeech/voc0](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/ljspeech/voc0).
### Parallel WaveGAN
[Parallel WaveGAN](https://arxiv.org/abs/1910.11480) trains a non-autoregressive WaveNet variant as a generator in a GAN based training method.
...
...
@@ -289,7 +289,7 @@ You can find PaddleSpeech TTS's WaveFlow with LJSpeech dataset example at [examp
You can find PaddleSpeech TTS's Parallel WaveGAN with CSMSC example at [examples/csmsc/voc1](https://github.com/PaddlePaddle/DeepSpeech/tree/develop/examples/csmsc/voc1).
You can find PaddleSpeech TTS's Parallel WaveGAN with CSMSC example at [examples/csmsc/voc1](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/csmsc/voc1).
@@ -18,7 +18,7 @@ The models in PaddleSpeech TTS have the following mapping relationship:
## Quick Start
Let's take a FastSpeech2 + Parallel WaveGAN with CSMSC dataset for instance. (./examples/csmsc/)(https://github.com/PaddlePaddle/DeepSpeech/tree/develop/examples/csmsc)
Let's take a FastSpeech2 + Parallel WaveGAN with CSMSC dataset for instance. (./examples/csmsc/)(https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/csmsc)
A TTS system mainly includes three modules: `Text Frontend`, `Acoustic model` and `Vocoder`. We provide a complete Chinese text frontend module in PaddleSpeech TTS, see exapmle in [examples/other/text_frontend/](https://github.com/PaddlePaddle/DeepSpeech/tree/develop/examples/other/text_frontend).
A TTS system mainly includes three modules: `Text Frontend`, `Acoustic model` and `Vocoder`. We provide a complete Chinese text frontend module in PaddleSpeech TTS, see exapmle in [examples/other/text_frontend/](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/other/text_frontend).
"Mel GAN |adversial loss <br> Multi-resolution STFT loss | adversial loss|\n",
"Multi-Band Mel GAN | adversial loss <br> full band Multi-resolution STFT loss <br> sub band Multi-resolution STFT loss |Multi-Scale Discriminator|\n",
"HiFi GAN |adversial loss <br> Feature Matching <br> Mel-Spectrogram Loss | Multi-Scale Discriminator <br> Multi-Period Discriminato|\n"
"HiFi GAN |adversial loss <br> Feature Matching <br> Mel-Spectrogram Loss | Multi-Scale Discriminator <br> Multi-Period Discriminator|\n"
This example contains code used to train a [Tacotron2 ](https://arxiv.org/abs/1712.05884) model with [AISHELL-3](http://www.aishelltech.com/aishell_3). The trained model can be used in Voice Cloning Task, We refer to the model structure of [Transfer Learning from Speaker Verification to Multispeaker Text-To-Speech Synthesis](https://arxiv.org/pdf/1806.04558.pdf) . The general steps are as follows:
1. Speaker Encoder: We use a Speaker Verification to train a speaker encoder. Datasets used in this task are different from those used in Tacotron2, because the transcriptions are not needed, we use more datasets, refer to [ge2e](https://github.com/PaddlePaddle/DeepSpeech/tree/develop/examples/other/ge2e).
1. Speaker Encoder: We use a Speaker Verification to train a speaker encoder. Datasets used in this task are different from those used in Tacotron2, because the transcriptions are not needed, we use more datasets, refer to [ge2e](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/other/ge2e).
2. Synthesizer: Then, we use the trained speaker encoder to generate utterance embedding for each sentence in AISHELL-3. This embedding is a extra input of Tacotron2 which will be concated with encoder outputs.
3. Vocoder: We use WaveFlow as the neural Vocoder, refer to [waveflow](https://github.com/PaddlePaddle/DeepSpeech/tree/develop/examples/ljspeech/voc0).
3. Vocoder: We use WaveFlow as the neural Vocoder, refer to [waveflow](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/ljspeech/voc0).
## Get Started
Assume the path to the dataset is `~/datasets/data_aishell3`.
...
...
@@ -39,9 +39,9 @@ There are silence in the edge of AISHELL-3's wavs, and the audio amplitude is ve
We use Montreal Force Aligner 1.0. The label in aishell3 include pinyin，so the lexicon we provided to MFA is pinyin rather than Chinese characters. And the prosody marks(`$` and `%`) need to be removed. You shoud preprocess the dataset into the format which MFA needs, the texts have the same name with wavs and have the suffix `.lab`.
We use [lexicon.txt](https://github.com/PaddlePaddle/DeepSpeech/blob/develop/paddlespeech/t2s/exps/voice_cloning/tacotron2_ge2e/lexicon.txt) as the lexicon.
We use [lexicon.txt](https://github.com/PaddlePaddle/PaddleSpeech/blob/develop/paddlespeech/t2s/exps/voice_cloning/tacotron2_ge2e/lexicon.txt) as the lexicon.
You can download the alignment results from here [alignment_aishell3.tar.gz](https://paddlespeech.bj.bcebos.com/Parakeet/alignment_aishell3.tar.gz), or train your own MFA model reference to [use_mfa example](https://github.com/PaddlePaddle/DeepSpeech/tree/develop/examples/other/use_mfa)(use MFA1.x now) of our repo.
You can download the alignment results from here [alignment_aishell3.tar.gz](https://paddlespeech.bj.bcebos.com/Parakeet/alignment_aishell3.tar.gz), or train your own MFA model reference to [use_mfa example](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/other/use_mfa)(use MFA1.x now) of our repo.
@@ -7,7 +7,7 @@ Download CSMSC from it's [Official Website](https://test.data-baker.com/data/ind
### Get MFA result of CSMSC and Extract it
We use [MFA](https://github.com/MontrealCorpusTools/Montreal-Forced-Aligner) to get durations for SPEEDYSPEECH.
You can download from here [baker_alignment_tone.tar.gz](https://paddlespeech.bj.bcebos.com/MFA/BZNSYP/with_tone/baker_alignment_tone.tar.gz), or train your own MFA model reference to [use_mfa example](https://github.com/PaddlePaddle/DeepSpeech/tree/develop/examples/other/use_mfa) of our repo.
You can download from here [baker_alignment_tone.tar.gz](https://paddlespeech.bj.bcebos.com/MFA/BZNSYP/with_tone/baker_alignment_tone.tar.gz), or train your own MFA model reference to [use_mfa example](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/other/use_mfa) of our repo.
## Get Started
Assume the path to the dataset is `~/datasets/BZNSYP`.
...
...
@@ -89,7 +89,7 @@ optional arguments:
6.`--tones-dict` is the path of the tone vocabulary file.
### Synthesize
We use [parallel wavegan](https://github.com/PaddlePaddle/DeepSpeech/tree/develop/examples/csmsc/voc1) as the neural vocoder.
We use [parallel wavegan](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/csmsc/voc1) as the neural vocoder.
Download pretrained parallel wavegan model from [pwg_baker_ckpt_0.4.zip](https://paddlespeech.bj.bcebos.com/Parakeet/pwg_baker_ckpt_0.4.zip) and unzip it.
@@ -7,7 +7,7 @@ Download CSMSC from it's [Official Website](https://test.data-baker.com/data/ind
### Get MFA result of CSMSC and Extract it
We use [MFA](https://github.com/MontrealCorpusTools/Montreal-Forced-Aligner) to get durations for fastspeech2.
You can download from here [baker_alignment_tone.tar.gz](https://paddlespeech.bj.bcebos.com/MFA/BZNSYP/with_tone/baker_alignment_tone.tar.gz), or train your own MFA model reference to [use_mfa example](https://github.com/PaddlePaddle/DeepSpeech/tree/develop/examples/other/use_mfa) of our repo.
You can download from here [baker_alignment_tone.tar.gz](https://paddlespeech.bj.bcebos.com/MFA/BZNSYP/with_tone/baker_alignment_tone.tar.gz), or train your own MFA model reference to [use_mfa example](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/other/use_mfa) of our repo.
## Get Started
Assume the path to the dataset is `~/datasets/BZNSYP`.
...
...
@@ -87,7 +87,7 @@ optional arguments:
5.`--phones-dict` is the path of the phone vocabulary file.
### Synthesize
We use [parallel wavegan](https://github.com/PaddlePaddle/DeepSpeech/tree/develop/examples/csmsc/voc1) as the neural vocoder.
We use [parallel wavegan](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/csmsc/voc1) as the neural vocoder.
Download pretrained parallel wavegan model from [pwg_baker_ckpt_0.4.zip](https://paddlespeech.bj.bcebos.com/Parakeet/pwg_baker_ckpt_0.4.zip) and unzip it.
@@ -6,7 +6,7 @@ Download CSMSC from the [official website](https://www.data-baker.com/data/index
### Get MFA results for silence trim
We use [MFA](https://github.com/MontrealCorpusTools/Montreal-Forced-Aligner) results to cut silence in the edge of audio.
You can download from here [baker_alignment_tone.tar.gz](https://paddlespeech.bj.bcebos.com/MFA/BZNSYP/with_tone/baker_alignment_tone.tar.gz), or train your own MFA model reference to [use_mfa example](https://github.com/PaddlePaddle/DeepSpeech/tree/develop/examples/other/use_mfa) of our repo.
You can download from here [baker_alignment_tone.tar.gz](https://paddlespeech.bj.bcebos.com/MFA/BZNSYP/with_tone/baker_alignment_tone.tar.gz), or train your own MFA model reference to [use_mfa example](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/other/use_mfa) of our repo.
## Get Started
Assume the path to the dataset is `~/datasets/BZNSYP`.
**Ps.** You can use [waveflow](https://github.com/PaddlePaddle/DeepSpeech/tree/develop/examples/ljspeech/voc0) as the neural vocoder to synthesize mels to wavs. (Please refer to `synthesize.sh` in our LJSpeech waveflow example)
**Ps.** You can use [waveflow](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/ljspeech/voc0) as the neural vocoder to synthesize mels to wavs. (Please refer to `synthesize.sh` in our LJSpeech waveflow example)
## Pretrained Models
Pretrained Models can be downloaded from links below. We provide 2 models with different configurations.
5.`--phones-dict` is the path of the phone vocabulary file.
## Synthesize
We use [waveflow](https://github.com/PaddlePaddle/DeepSpeech/tree/develop/examples/ljspeech/voc0) as the neural vocoder.
We use [waveflow](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/ljspeech/voc0) as the neural vocoder.
Download Pretrained WaveFlow Model with residual channel equals 128 from [waveflow_ljspeech_ckpt_0.3.zip](https://paddlespeech.bj.bcebos.com/Parakeet/waveflow_ljspeech_ckpt_0.3.zip) and unzip it.
@@ -7,7 +7,7 @@ Download LJSpeech-1.1 from the [official website](https://keithito.com/LJ-Speech
### Get MFA result of LJSpeech-1.1 and Extract it
We use [MFA](https://github.com/MontrealCorpusTools/Montreal-Forced-Aligner) to get durations for fastspeech2.
You can download from here [ljspeech_alignment.tar.gz](https://paddlespeech.bj.bcebos.com/MFA/LJSpeech-1.1/ljspeech_alignment.tar.gz), or train your own MFA model reference to [use_mfa example](https://github.com/PaddlePaddle/DeepSpeech/tree/develop/examples/other/use_mfa) of our repo.
You can download from here [ljspeech_alignment.tar.gz](https://paddlespeech.bj.bcebos.com/MFA/LJSpeech-1.1/ljspeech_alignment.tar.gz), or train your own MFA model reference to [use_mfa example](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/other/use_mfa) of our repo.
## Get Started
Assume the path to the dataset is `~/datasets/LJSpeech-1.1`.
...
...
@@ -86,7 +86,7 @@ optional arguments:
5.`--phones-dict` is the path of the phone vocabulary file.
### Synthesize
We use [parallel wavegan](https://github.com/PaddlePaddle/DeepSpeech/tree/develop/examples/ljspeech/voc1) as the neural vocoder.
We use [parallel wavegan](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/ljspeech/voc1) as the neural vocoder.
Download pretrained parallel wavegan model from [pwg_ljspeech_ckpt_0.5.zip](https://paddlespeech.bj.bcebos.com/Parakeet/pwg_ljspeech_ckpt_0.5.zip) and unzip it.
@@ -5,7 +5,7 @@ This example contains code used to train a [parallel wavegan](http://arxiv.org/a
Download LJSpeech-1.1 from the [official website](https://keithito.com/LJ-Speech-Dataset/).
### Get MFA results for silence trim
We use [MFA](https://github.com/MontrealCorpusTools/Montreal-Forced-Aligner) results to cut silence in the edge of audio.
You can download from here [ljspeech_alignment.tar.gz](https://paddlespeech.bj.bcebos.com/MFA/LJSpeech-1.1/ljspeech_alignment.tar.gz), or train your own MFA model reference to [use_mfa example](https://github.com/PaddlePaddle/DeepSpeech/tree/develop/examples/other/use_mfa) of our repo.
You can download from here [ljspeech_alignment.tar.gz](https://paddlespeech.bj.bcebos.com/MFA/LJSpeech-1.1/ljspeech_alignment.tar.gz), or train your own MFA model reference to [use_mfa example](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/other/use_mfa) of our repo.
## Get Started
Assume the path to the dataset is `~/datasets/LJSpeech-1.1`.
This experiment trains a speaker encoder with speaker verification as its task. It is done as a part of the experiment of transfer learning from speaker verification to multispeaker text-to-speech synthesis, which can be found at [examples/aishell3/vc0](https://github.com/PaddlePaddle/DeepSpeech/tree/develop/examples/aishell3/vc0). The trained speaker encoder is used to extract utterance embeddings from utterances.
This experiment trains a speaker encoder with speaker verification as its task. It is done as a part of the experiment of transfer learning from speaker verification to multispeaker text-to-speech synthesis, which can be found at [examples/aishell3/vc0](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/aishell3/vc0). The trained speaker encoder is used to extract utterance embeddings from utterances.
## Model
The model used in this experiment is the speaker encoder with text independent speaker verification task in [GENERALIZED END-TO-END LOSS FOR SPEAKER VERIFICATION](https://arxiv.org/pdf/1710.10467.pdf). GE2E-softmax loss is used.
@@ -7,8 +7,8 @@ Download VCTK-0.92 from the [official website](https://datashare.ed.ac.uk/handle
### Get MFA result of VCTK and Extract it
We use [MFA](https://github.com/MontrealCorpusTools/Montreal-Forced-Aligner) to get durations for fastspeech2.
You can download from here [vctk_alignment.tar.gz](https://paddlespeech.bj.bcebos.com/MFA/VCTK-Corpus-0.92/vctk_alignment.tar.gz), or train your own MFA model reference to [use_mfa example](https://github.com/PaddlePaddle/DeepSpeech/tree/develop/examples/other/use_mfa) of our repo.
ps: we remove three speakers in VCTK-0.92 (see [reorganize_vctk.py](https://github.com/PaddlePaddle/DeepSpeech/blob/develop/examples/other/use_mfa/local/reorganize_vctk.py)):
You can download from here [vctk_alignment.tar.gz](https://paddlespeech.bj.bcebos.com/MFA/VCTK-Corpus-0.92/vctk_alignment.tar.gz), or train your own MFA model reference to [use_mfa example](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/other/use_mfa) of our repo.
ps: we remove three speakers in VCTK-0.92 (see [reorganize_vctk.py](https://github.com/PaddlePaddle/PaddleSpeech/blob/develop/examples/other/use_mfa/local/reorganize_vctk.py)):
1.`p315`, because no txt for it.
2.`p280` and `p362`, because no *_mic2.flac (which is better than *_mic1.flac) for them.
...
...
@@ -88,7 +88,7 @@ optional arguments:
4.`--phones-dict` is the path of the phone vocabulary file.
### Synthesize
We use [parallel wavegan](https://github.com/PaddlePaddle/DeepSpeech/tree/develop/examples/vctk/voc1) as the neural vocoder.
We use [parallel wavegan](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/vctk/voc1) as the neural vocoder.
Download pretrained parallel wavegan model from [pwg_vctk_ckpt_0.5.zip](https://paddlespeech.bj.bcebos.com/Parakeet/pwg_vctk_ckpt_0.5.zip)and unzip it.
@@ -7,8 +7,8 @@ Download VCTK-0.92 from the [official website](https://datashare.ed.ac.uk/handl
### Get MFA results for silence trim
We use [MFA](https://github.com/MontrealCorpusTools/Montreal-Forced-Aligner) results to cut silence in the edge of audio.
You can download from here [vctk_alignment.tar.gz](https://paddlespeech.bj.bcebos.com/MFA/VCTK-Corpus-0.92/vctk_alignment.tar.gz), or train your own MFA model reference to [use_mfa example](https://github.com/PaddlePaddle/DeepSpeech/tree/develop/examples/other/use_mfa) of our repo.
ps: we remove three speakers in VCTK-0.92 (see [reorganize_vctk.py](https://github.com/PaddlePaddle/DeepSpeech/blob/develop/examples/other/use_mfa/local/reorganize_vctk.py)):
You can download from here [vctk_alignment.tar.gz](https://paddlespeech.bj.bcebos.com/MFA/VCTK-Corpus-0.92/vctk_alignment.tar.gz), or train your own MFA model reference to [use_mfa example](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/other/use_mfa) of our repo.
ps: we remove three speakers in VCTK-0.92 (see [reorganize_vctk.py](https://github.com/PaddlePaddle/PaddleSpeech/blob/develop/examples/other/use_mfa/local/reorganize_vctk.py)):
1.`p315`, because no txt for it.
2.`p280` and `p362`, because no *_mic2.flac (which is better than *_mic1.flac) for them.
