RyanSpeech model (based on ESPnet2)
espnet/english_male_ryanspeech_fastspeech
This model was trained by Rohola Zandie using ryanspeech recipe in espnet. For the best results you need to download the vocoder separately from here and then use the following code:
from espnet2.bin.tts_inference import Text2Speech
from scipy.io.wavfile import write
model = Text2Speech.from_pretrained(
model_file="espnet/english_male_ryanspeech_fastspeech",
vocoder_file="path_to_vocoder/train_nodev_parallel_wavegan.v1.long/checkpoint-1000000steps.pkl"
)
output = model("This is a simple test.")
write("x.wav", 22050, output['wav'].numpy())
Download the dataset
You can download RyanSpeech dataset from here or here.
TTS config
<details><summary>expand</summary>
config: conf/tuning/train_fastspeech.yaml
print_config: false
log_level: INFO
dry_run: false
iterator_type: sequence
output_dir: exp/tts_train_fastspeech_raw_phn_tacotron_g2p_en_no_space
ngpu: 1
seed: 0
num_workers: 1
num_att_plot: 3
dist_backend: nccl
dist_init_method: env://
dist_world_size: null
dist_rank: null
local_rank: 0
dist_master_addr: null
dist_master_port: null
dist_launcher: null
multiprocessing_distributed: false
cudnn_enabled: true
cudnn_benchmark: false
cudnn_deterministic: true
collect_stats: false
write_collected_feats: false
max_epoch: 1000
patience: null
val_scheduler_criterion:
- valid
- loss
early_stopping_criterion:
- valid
- loss
- min
best_model_criterion:
- - valid
- loss
- min
- - train
- loss
- min
keep_nbest_models: 5
grad_clip: 1.0
grad_clip_type: 2.0
grad_noise: false
accum_grad: 6
no_forward_run: false
resume: true
train_dtype: float32
use_amp: false
log_interval: null
pretrain_path: []
pretrain_key: []
num_iters_per_epoch: 500
batch_size: 20
valid_batch_size: null
batch_bins: 800000
valid_batch_bins: null
train_shape_file:
- exp/tts_train_raw_phn_tacotron_g2p_en_no_space/decode_use_teacher_forcingtrue_train.loss.best/stats/train/text_shape.phn
- exp/tts_train_raw_phn_tacotron_g2p_en_no_space/decode_use_teacher_forcingtrue_train.loss.best/stats/train/speech_shape
valid_shape_file:
- exp/tts_train_raw_phn_tacotron_g2p_en_no_space/decode_use_teacher_forcingtrue_train.loss.best/stats/valid/text_shape.phn
- exp/tts_train_raw_phn_tacotron_g2p_en_no_space/decode_use_teacher_forcingtrue_train.loss.best/stats/valid/speech_shape
batch_type: numel
valid_batch_type: null
fold_length:
- 150
- 204800
sort_in_batch: descending
sort_batch: descending
multiple_iterator: false
chunk_length: 500
chunk_shift_ratio: 0.5
num_cache_chunks: 1024
train_data_path_and_name_and_type:
- - dump/raw/tr_no_dev/text
- text
- text
- - exp/tts_train_raw_phn_tacotron_g2p_en_no_space/decode_use_teacher_forcingtrue_train.loss.best//tr_no_dev/durations
- durations
- text_int
- - dump/raw/tr_no_dev/wav.scp
- speech
- sound
valid_data_path_and_name_and_type:
- - dump/raw/dev/text
- text
- text
- - exp/tts_train_raw_phn_tacotron_g2p_en_no_space/decode_use_teacher_forcingtrue_train.loss.best//dev/durations
- durations
- text_int
- - dump/raw/dev/wav.scp
- speech
- sound
allow_variable_data_keys: false
max_cache_size: 0.0
max_cache_fd: 32
valid_max_cache_size: null
optim: adam
optim_conf:
lr: 1.0
scheduler: noamlr
scheduler_conf:
model_size: 384
warmup_steps: 4000
token_list:
- <blank>
- <unk>
- AH0
- T
- N
- S
- R
- D
- L
- K
- IH1
- M
- EH1
- Z
- DH
- UW1
- AE1
- IH0
- AY1
- AH1
- W
- .
- P
- F
- IY1
- V
- ER0
- AA1
- B
- AO1
- HH
- EY1
- IY0
- ','
- Y
- NG
- OW1
- G
- AW1
- TH
- SH
- UH1
- '?'
- ER1
- JH
- CH
- OW0
- OW2
- EH2
- IH2
- EY2
- AA2
- AE2
- AY2
- ''''
- OY1
- UW0
- '!'
- AO2
- EH0
- ZH
- AH2
- AE0
- UW2
- AA0
- AY0
- IY2
- AW2
- AO0
- EY0
- ER2
- UH2
- '...'
- AW0
- UH0
- OY2
- <sos/eos>
odim: null
model_conf: {}
use_preprocessor: true
token_type: phn
bpemodel: null
non_linguistic_symbols: null
cleaner: tacotron
g2p: g2p_en_no_space
feats_extract: fbank
feats_extract_conf:
fs: 22050
fmin: 80
fmax: 7600
n_mels: 80
hop_length: 256
n_fft: 1024
win_length: null
normalize: global_mvn
normalize_conf:
stats_file: exp/tts_train_raw_phn_tacotron_g2p_en_no_space/decode_use_teacher_forcingtrue_train.loss.best/stats/train/feats_stats.npz
tts: fastspeech
tts_conf:
adim: 384
aheads: 2
elayers: 6
eunits: 1536
dlayers: 6
dunits: 1536
positionwise_layer_type: conv1d
positionwise_conv_kernel_size: 3
duration_predictor_layers: 2
duration_predictor_chans: 384
duration_predictor_kernel_size: 3
postnet_layers: 5
postnet_filts: 5
postnet_chans: 256
use_masking: true
use_scaled_pos_enc: true
encoder_normalize_before: true
decoder_normalize_before: true
reduction_factor: 1
init_type: xavier_uniform
init_enc_alpha: 1.0
init_dec_alpha: 1.0
transformer_enc_dropout_rate: 0.1
transformer_enc_positional_dropout_rate: 0.1
transformer_enc_attn_dropout_rate: 0.1
transformer_dec_dropout_rate: 0.1
transformer_dec_positional_dropout_rate: 0.1
transformer_dec_attn_dropout_rate: 0.1
pitch_extract: null
pitch_extract_conf: {}
pitch_normalize: null
pitch_normalize_conf: {}
energy_extract: null
energy_extract_conf: {}
energy_normalize: null
energy_normalize_conf: {}
required:
- output_dir
- token_list
distributed: false
</details>
Citing RyanSpeech
@inproceedings{Zandie2021RyanSpeechAC,
title={RyanSpeech: A Corpus for Conversational Text-to-Speech Synthesis},
author={Rohola Zandie and Mohammad H. Mahoor and Julia Madsen and Eshrat S. Emamian},
booktitle={Interspeech},
year={2021}
}