metadata

language: en
tags:
  - text-to-speech
  - TTS
  - speech-synthesis
  - fastspeech2
  - speechbrain
license: apache-2.0
datasets:
  - LJSpeech
metrics:
  - mos

Text-to-Speech (TTS) with FastSpeech2-Internal-Alignment trained on LJSpeech

This repository provides all the necessary tools for Text-to-Speech (TTS) with SpeechBrain using a FastSpeech2 with internal alignment pretrained on LJSpeech.

The pre-trained model takes texts or phonemes as input and produces a spectrogram in output. One can get the final waveform by applying a vocoder (e.g., HiFIGAN) on top of the generated spectrogram. It should be noted that if the input is text, we use a state-of-the-art grapheme-to-phoneme module to convert it to phonemes and then pass the phonemes to fastspeech2-internal-alignment model.

Install SpeechBrain

git clone https://github.com/speechbrain/speechbrain.git
cd speechbrain
pip install -r requirements.txt
pip install --editable .

Please notice that we encourage you to read our tutorials and learn more about SpeechBrain.

Perform Text-to-Speech (TTS) with FastSpeech2-Internal-Alignment

import torchaudio
from speechbrain.pretrained import FastSpeech2InternalAlignment
from speechbrain.pretrained import HIFIGAN

# Intialize TTS (tacotron2) and Vocoder (HiFIGAN)
fastspeech2 = FastSpeech2InternalAlignment.from_hparams(source="speechbrain/tts-fastspeech2-internal-alignment-ljspeech", savedir="tmpdir_tts")
hifi_gan = HIFIGAN.from_hparams(source="speechbrain/tts-hifigan-ljspeech", savedir="tmpdir_vocoder")

# Run TTS with text input
input_text = "Welcome to speechbrain, this is a test run with fastspeech internal alignment."

mel_output, durations, pitch, energy = fastspeech2.encode_text(
  [input_text],
  pace=1.0,        # scale up/down the speed
  pitch_rate=1.0,  # scale up/down the pitch
  energy_rate=1.0, # scale up/down the energy
)

# Running Vocoder (spectrogram-to-waveform)
waveforms = hifi_gan.decode_batch(mel_output)

# Save the waverform
torchaudio.save('example_TTS_input_text.wav', waveforms.squeeze(1), 22050)


# Run TTS with phoneme input
input_phonemes = ['W', 'ER', ' ', 'DH', 'AH', ' ', 'L', 'IY', 'D', 'ER', 'Z', ' ', 'IH', 'N', ' ', 'DH', 'IH', 'S', ' ', 'L', 'AH', 'K', 'L', 'AH', 'S', ' ', 'CH', 'EY', 'N', 'JH', ';', " ", 'DH', 'OW', ' ', 'AW', 'ER', ' ', 'OW', 'N', ' ', 'B', 'AE', 'S', 'K', 'ER', 'V', 'IH', 'L', ';', " ", 'HH', 'UW', ' ', 'W', 'AA', 'Z', ' ', 'AE', 'T', ' ', 'W', 'ER', 'K', ' ', 'S', 'AH', 'M', ' ', 'Y', 'IH', 'R', 'Z', ' ', 'B', 'IH', 'F', 'AO', 'R', ' ', 'DH', 'EH', 'M', ';', " ", 'W', 'EH', 'N', 'T', ' ', 'M', 'AH', 'CH', ' ', 'AA', 'N', ' ', 'DH', 'AH', ' ', 'S', 'EY', 'M', ' ', 'L', 'AY', 'N', 'Z', ';']

mel_output, durations, pitch, energy = fastspeech2.encode_phoneme(
  [input_phonemes],
  pace=1.0,        # scale up/down the speed
  pitch_rate=1.0,  # scale up/down the pitch
  energy_rate=1.0, # scale up/down the energy
)

# Running Vocoder (spectrogram-to-waveform)
waveforms = hifi_gan.decode_batch(mel_output)

# Save the waverform
torchaudio.save('example_TTS_input_phoneme.wav', waveforms.squeeze(1), 22050)

If you want to generate multiple sentences in one-shot, you can do in this way:

from speechbrain.pretrained import FastSpeech2InternalAlignment
fastspeech2 = FastSpeech2InternalAlignment.from_hparams(source="speechbrain/tts-fastspeech2-internal-alignment-ljspeech", savedir="tmpdir_tts")
items = [
       "A quick brown fox jumped over the lazy dog",
       "How much wood would a woodchuck chuck?",
       "Never odd or even"
]
mel_outputs, durations, pitch, energy = fastspeech2.encode_text(
  items,
  pace=1.0,        # scale up/down the speed
  pitch_rate=1.0,  # scale up/down the pitch
  energy_rate=1.0, # scale up/down the energy
)

Inference on GPU

To perform inference on the GPU, add run_opts={"device":"cuda"} when calling the from_hparams method.

Training

The model was trained with SpeechBrain. To train it from scratch follow these steps:

Clone SpeechBrain:

git clone https://github.com/speechbrain/speechbrain/

Install it:

cd speechbrain
pip install -r requirements.txt
pip install -e .

Run Training:

cd recipes/LJSpeech/TTS/fastspeech2/
python train_internal_alignment.py hparams/train_internal_alignment.yaml --data_folder=/your_folder/LJSpeech-1.1

You can find our training results (models, logs, etc) here.

Limitations

The SpeechBrain team does not provide any warranty on the performance achieved by this model when used on other datasets.

About SpeechBrain

Website: https://speechbrain.github.io/
Code: https://github.com/speechbrain/speechbrain/
HuggingFace: https://huggingface.co/speechbrain/

Citing SpeechBrain

Please, cite SpeechBrain if you use it for your research or business.

@misc{speechbrain,
  title={{SpeechBrain}: A General-Purpose Speech Toolkit},
  author={Mirco Ravanelli and Titouan Parcollet and Peter Plantinga and Aku Rouhe and Samuele Cornell and Loren Lugosch and Cem Subakan and Nauman Dawalatabad and Abdelwahab Heba and Jianyuan Zhong and Ju-Chieh Chou and Sung-Lin Yeh and Szu-Wei Fu and Chien-Feng Liao and Elena Rastorgueva and François Grondin and William Aris and Hwidong Na and Yan Gao and Renato De Mori and Yoshua Bengio},
  year={2021},
  eprint={2106.04624},
  archivePrefix={arXiv},
  primaryClass={eess.AS},
  note={arXiv:2106.04624}
}