Upload full bundle: fp16 + fp32 models, vocoder, samples, code, licenses
Browse files- .gitattributes +5 -0
- LICENSE +34 -0
- NOTICE +34 -0
- api.py +102 -0
- burmese.py +42 -0
- config.json +35 -0
- model_fp16.pt +3 -0
- model_fp32.pt +3 -0
- samples/sample_0.wav +3 -0
- samples/sample_1.wav +3 -0
- samples/sample_2.wav +3 -0
- samples/sample_3.wav +3 -0
- samples/sample_4.wav +3 -0
- symbols.py +57 -0
- transcripts.json +194 -0
- vocab.txt +98 -0
- vocos.pt +3 -0
.gitattributes
CHANGED
|
@@ -33,3 +33,8 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
|
|
| 33 |
*.zip filter=lfs diff=lfs merge=lfs -text
|
| 34 |
*.zst filter=lfs diff=lfs merge=lfs -text
|
| 35 |
*tfevents* filter=lfs diff=lfs merge=lfs -text
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 33 |
*.zip filter=lfs diff=lfs merge=lfs -text
|
| 34 |
*.zst filter=lfs diff=lfs merge=lfs -text
|
| 35 |
*tfevents* filter=lfs diff=lfs merge=lfs -text
|
| 36 |
+
samples/sample_0.wav filter=lfs diff=lfs merge=lfs -text
|
| 37 |
+
samples/sample_1.wav filter=lfs diff=lfs merge=lfs -text
|
| 38 |
+
samples/sample_2.wav filter=lfs diff=lfs merge=lfs -text
|
| 39 |
+
samples/sample_3.wav filter=lfs diff=lfs merge=lfs -text
|
| 40 |
+
samples/sample_4.wav filter=lfs diff=lfs merge=lfs -text
|
LICENSE
ADDED
|
@@ -0,0 +1,34 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
Creative Commons Legal Code
|
| 2 |
+
|
| 3 |
+
CC0 1.0 Universal
|
| 4 |
+
|
| 5 |
+
CREATIVE COMMONS CORPORATION IS NOT A LAW FIRM AND DOES NOT PROVIDE
|
| 6 |
+
LEGAL SERVICES. COPYING, DISTRIBUTING, OR OTHERWISE USING THIS WORK
|
| 7 |
+
IS DONE AT YOUR OWN RISK.
|
| 8 |
+
|
| 9 |
+
Statement of Purpose
|
| 10 |
+
|
| 11 |
+
The laws of most jurisdictions throughout the world automatically confer
|
| 12 |
+
exclusive Copyright and Related Rights (defined below) upon the creator
|
| 13 |
+
and subsequent owner(s) (each and all, an "owner") of an original work
|
| 14 |
+
of authorship and/or a database (each, a "Work").
|
| 15 |
+
|
| 16 |
+
Certain owners wish to permanently relinquish those rights to a Work for
|
| 17 |
+
the purpose of contributing to a commons of creative, cultural and
|
| 18 |
+
scientific works ("Commons") that the public can reliably and without
|
| 19 |
+
fear of later claims of infringement build upon, modify, incorporate in
|
| 20 |
+
other works, reuse and redistribute as freely as possible in any form
|
| 21 |
+
whatsoever and for any purposes, including without limitation commercial
|
| 22 |
+
purposes.
|
| 23 |
+
|
| 24 |
+
To the greatest extent permitted by, but not in contravention of,
|
| 25 |
+
applicable law, Affirmer hereby overtly, fully, permanently, irrevocably
|
| 26 |
+
and unconditionally waives, abandons, and surrenders all of Affirmer's
|
| 27 |
+
Copyright and Related Rights and associated claims and causes of action,
|
| 28 |
+
whether now known or unknown (including existing as well as future claims
|
| 29 |
+
and causes of action), in the Work.
|
| 30 |
+
|
| 31 |
+
You can copy, modify, distribute and perform the Work, even for commercial
|
| 32 |
+
purposes, all without asking permission.
|
| 33 |
+
|
| 34 |
+
For full legal text: https://creativecommons.org/publicdomain/zero/1.0/legalcode
|
NOTICE
ADDED
|
@@ -0,0 +1,34 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# NOTICE
|
| 2 |
+
|
| 3 |
+
## Model Weights (CC0 1.0 Universal)
|
| 4 |
+
- `model_fp16.pt`
|
| 5 |
+
- `model_fp32.pt`
|
| 6 |
+
|
| 7 |
+
Trained from scratch by freococo. Released to the public domain under CC0 1.0.
|
| 8 |
+
|
| 9 |
+
## Model Outputs (CC0 1.0 Universal)
|
| 10 |
+
- `samples/*.wav`
|
| 11 |
+
- Any audio generated by this model
|
| 12 |
+
|
| 13 |
+
Generated by the model above. Public domain.
|
| 14 |
+
|
| 15 |
+
## Adapted Code (MIT License)
|
| 16 |
+
- `api.py`
|
| 17 |
+
- `symbols.py`
|
| 18 |
+
- `burmese.py`
|
| 19 |
+
|
| 20 |
+
These files are derived from StableTTS (https://github.com/KdaiP/StableTTS),
|
| 21 |
+
which is distributed under the MIT License. The original MIT terms apply.
|
| 22 |
+
|
| 23 |
+
## Vocoder (MIT License)
|
| 24 |
+
- `vocos.pt`
|
| 25 |
+
|
| 26 |
+
Pretrained Vocos vocoder from the StableTTS1.1 release
|
| 27 |
+
(https://huggingface.co/KdaiP/StableTTS1.1), distributed under the MIT License.
|
| 28 |
+
Not modified by freococo โ used as-is.
|
| 29 |
+
|
| 30 |
+
## Summary
|
| 31 |
+
|
| 32 |
+
If you use this model, you're free to use the weights and outputs under CC0
|
| 33 |
+
(no attribution required). The supporting code and vocoder retain their
|
| 34 |
+
original MIT licenses.
|
api.py
ADDED
|
@@ -0,0 +1,102 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import torch
|
| 2 |
+
import torch.nn as nn
|
| 3 |
+
|
| 4 |
+
from dataclasses import asdict
|
| 5 |
+
|
| 6 |
+
from utils.audio import LogMelSpectrogram
|
| 7 |
+
from config import ModelConfig, MelConfig
|
| 8 |
+
from models.model import StableTTS
|
| 9 |
+
|
| 10 |
+
from text import symbols
|
| 11 |
+
from text import cleaned_text_to_sequence
|
| 12 |
+
from text.mandarin import chinese_to_cnm3
|
| 13 |
+
from text.english import english_to_ipa2
|
| 14 |
+
from text.japanese import japanese_to_ipa2
|
| 15 |
+
|
| 16 |
+
from datas.dataset import intersperse
|
| 17 |
+
from utils.audio import load_and_resample_audio
|
| 18 |
+
|
| 19 |
+
def get_vocoder(model_path, model_name='ffgan') -> nn.Module:
|
| 20 |
+
if model_name == 'ffgan':
|
| 21 |
+
# training or changing ffgan config is not supported in this repo
|
| 22 |
+
# you can train your own model at https://github.com/fishaudio/vocoder
|
| 23 |
+
from vocoders.ffgan.model import FireflyGANBaseWrapper
|
| 24 |
+
vocoder = FireflyGANBaseWrapper(model_path)
|
| 25 |
+
|
| 26 |
+
elif model_name == 'vocos':
|
| 27 |
+
from vocoders.vocos.models.model import Vocos
|
| 28 |
+
from config import VocosConfig, MelConfig
|
| 29 |
+
vocoder = Vocos(VocosConfig(), MelConfig())
|
| 30 |
+
vocoder.load_state_dict(torch.load(model_path, weights_only=True, map_location='cpu'))
|
| 31 |
+
vocoder.eval()
|
| 32 |
+
|
| 33 |
+
else:
|
| 34 |
+
raise NotImplementedError(f"Unsupported model: {model_name}")
|
| 35 |
+
|
| 36 |
+
return vocoder
|
| 37 |
+
|
| 38 |
+
class StableTTSAPI(nn.Module):
|
| 39 |
+
def __init__(self, tts_model_path, vocoder_model_path, vocoder_name='ffgan'):
|
| 40 |
+
super().__init__()
|
| 41 |
+
|
| 42 |
+
self.mel_config = MelConfig()
|
| 43 |
+
self.tts_model_config = ModelConfig()
|
| 44 |
+
|
| 45 |
+
self.mel_extractor = LogMelSpectrogram(**asdict(self.mel_config))
|
| 46 |
+
|
| 47 |
+
# text to mel spectrogram
|
| 48 |
+
self.tts_model = StableTTS(len(symbols), self.mel_config.n_mels, **asdict(self.tts_model_config))
|
| 49 |
+
self.tts_model.load_state_dict(torch.load(tts_model_path, map_location='cpu', weights_only=True))
|
| 50 |
+
self.tts_model.eval()
|
| 51 |
+
|
| 52 |
+
# mel spectrogram to waveform
|
| 53 |
+
self.vocoder_model = get_vocoder(vocoder_model_path, vocoder_name)
|
| 54 |
+
self.vocoder_model.eval()
|
| 55 |
+
|
| 56 |
+
self.g2p_mapping = {
|
| 57 |
+
'chinese': chinese_to_cnm3,
|
| 58 |
+
'japanese': japanese_to_ipa2,
|
| 59 |
+
'english': english_to_ipa2,
|
| 60 |
+
}
|
| 61 |
+
self.supported_languages = self.g2p_mapping.keys()
|
| 62 |
+
|
| 63 |
+
@ torch.inference_mode()
|
| 64 |
+
def inference(self, text, ref_audio, language, step, temperature=1.0, length_scale=1.0, solver=None, cfg=3.0):
|
| 65 |
+
device = next(self.parameters()).device
|
| 66 |
+
phonemizer = self.g2p_mapping.get(language)
|
| 67 |
+
|
| 68 |
+
text = phonemizer(text)
|
| 69 |
+
text = torch.tensor(intersperse(cleaned_text_to_sequence(text), item=0), dtype=torch.long, device=device).unsqueeze(0)
|
| 70 |
+
text_length = torch.tensor([text.size(-1)], dtype=torch.long, device=device)
|
| 71 |
+
|
| 72 |
+
ref_audio = load_and_resample_audio(ref_audio, self.mel_config.sample_rate).to(device)
|
| 73 |
+
ref_audio = self.mel_extractor(ref_audio)
|
| 74 |
+
|
| 75 |
+
mel_output = self.tts_model.synthesise(text, text_length, step, temperature, ref_audio, length_scale, solver, cfg)['decoder_outputs']
|
| 76 |
+
audio_output = self.vocoder_model(mel_output)
|
| 77 |
+
return audio_output.cpu(), mel_output.cpu()
|
| 78 |
+
|
| 79 |
+
def get_params(self):
|
| 80 |
+
tts_param = sum(p.numel() for p in self.tts_model.parameters()) / 1e6
|
| 81 |
+
vocoder_param = sum(p.numel() for p in self.vocoder_model.parameters()) / 1e6
|
| 82 |
+
return tts_param, vocoder_param
|
| 83 |
+
|
| 84 |
+
if __name__ == '__main__':
|
| 85 |
+
device = 'cuda' if torch.cuda.is_available() else 'cpu'
|
| 86 |
+
tts_model_path = './checkpoints/checkpoint_0.pt'
|
| 87 |
+
vocoder_model_path = './vocoders/pretrained/vocos.pt'
|
| 88 |
+
|
| 89 |
+
model = StableTTSAPI(tts_model_path, vocoder_model_path, 'vocos')
|
| 90 |
+
model.to(device)
|
| 91 |
+
|
| 92 |
+
text = 'ๆจฑ่ฝๆปกๆฎ็ฅๅฟต้โฆโฆๆฎๆญ่ฑ่ฝ้ๆ็ฅโฆโฆๆจฑ่ฑๆปกๅฐ้ไบๆๅฟโฆโฆๆฒ่็บท้ฃ็ฅๆฟ็ธ้โฆโฆ'
|
| 93 |
+
audio = './audio_1.wav'
|
| 94 |
+
|
| 95 |
+
audio_output, mel_output = model.inference(text, audio, 'chinese', 10, solver='dopri5', cfg=3)
|
| 96 |
+
print(audio_output.shape)
|
| 97 |
+
print(mel_output.shape)
|
| 98 |
+
|
| 99 |
+
import torchaudio
|
| 100 |
+
torchaudio.save('output.wav', audio_output, MelConfig().sample_rate)
|
| 101 |
+
|
| 102 |
+
|
burmese.py
ADDED
|
@@ -0,0 +1,42 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
"""
|
| 2 |
+
Character-level text frontend for Burmese.
|
| 3 |
+
No real G2P: we clean the text and pass characters through directly.
|
| 4 |
+
Matches the interface of english_to_ipa2 / japanese_to_ipa2 (returns a list).
|
| 5 |
+
"""
|
| 6 |
+
import re
|
| 7 |
+
|
| 8 |
+
# Burmese Unicode ranges
|
| 9 |
+
_BURMESE_RANGES = (
|
| 10 |
+
(0x1000, 0x109F), # Myanmar
|
| 11 |
+
(0xAA60, 0xAA7F), # Myanmar Extended-A
|
| 12 |
+
(0xA9E0, 0xA9FF), # Myanmar Extended-B
|
| 13 |
+
)
|
| 14 |
+
|
| 15 |
+
# Allow these through the cleaner; everything else gets dropped
|
| 16 |
+
_ALLOWED_ASCII_PUNCT = set(".,!?'\"-:;()")
|
| 17 |
+
_ALLOWED_BURMESE_PUNCT = set("\u104A\u104B\u104C\u104D\u104E\u104F")
|
| 18 |
+
|
| 19 |
+
|
| 20 |
+
def _is_burmese(ch):
|
| 21 |
+
o = ord(ch)
|
| 22 |
+
return any(lo <= o <= hi for lo, hi in _BURMESE_RANGES)
|
| 23 |
+
|
| 24 |
+
|
| 25 |
+
def clean_burmese(text):
|
| 26 |
+
"""Keep only Burmese chars, spaces, and whitelisted punctuation."""
|
| 27 |
+
out = []
|
| 28 |
+
for ch in text:
|
| 29 |
+
if ch in (" ", "\t", "\n", "\r"):
|
| 30 |
+
out.append(" ")
|
| 31 |
+
elif _is_burmese(ch):
|
| 32 |
+
out.append(ch)
|
| 33 |
+
elif ch in _ALLOWED_ASCII_PUNCT or ch in _ALLOWED_BURMESE_PUNCT:
|
| 34 |
+
out.append(ch)
|
| 35 |
+
# else: drop silently
|
| 36 |
+
return re.sub(r"\s+", " ", "".join(out)).strip()
|
| 37 |
+
|
| 38 |
+
|
| 39 |
+
def burmese_to_ipa2(text):
|
| 40 |
+
"""Return a list of character tokens (matches the IPA-G2P interface)."""
|
| 41 |
+
cleaned = clean_burmese(text)
|
| 42 |
+
return list(cleaned)
|
config.json
ADDED
|
@@ -0,0 +1,35 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"mel_config": {
|
| 3 |
+
"sample_rate": 44100,
|
| 4 |
+
"n_fft": 2048,
|
| 5 |
+
"win_length": 2048,
|
| 6 |
+
"hop_length": 512,
|
| 7 |
+
"f_min": 0.0,
|
| 8 |
+
"f_max": null,
|
| 9 |
+
"pad": 768,
|
| 10 |
+
"n_mels": 128,
|
| 11 |
+
"center": false,
|
| 12 |
+
"pad_mode": "reflect",
|
| 13 |
+
"mel_scale": "slaney"
|
| 14 |
+
},
|
| 15 |
+
"model_config": {
|
| 16 |
+
"hidden_channels": 256,
|
| 17 |
+
"filter_channels": 1024,
|
| 18 |
+
"n_heads": 4,
|
| 19 |
+
"n_enc_layers": 3,
|
| 20 |
+
"n_dec_layers": 6,
|
| 21 |
+
"kernel_size": 3,
|
| 22 |
+
"p_dropout": 0.1,
|
| 23 |
+
"gin_channels": 256
|
| 24 |
+
},
|
| 25 |
+
"vocab_size": 98,
|
| 26 |
+
"sample_rate": 44100,
|
| 27 |
+
"n_mels": 128,
|
| 28 |
+
"hop_length": 512,
|
| 29 |
+
"n_fft": 2048,
|
| 30 |
+
"language": "burmese",
|
| 31 |
+
"tokenizer": "character_level",
|
| 32 |
+
"vocoder": "vocos",
|
| 33 |
+
"repo_ckpt": "freococo/StableTTS-Burmese-CKPT",
|
| 34 |
+
"repo_code": "freococo/StableTTS-Burmese-CODE"
|
| 35 |
+
}
|
model_fp16.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:48c778d162141ed399a17be67c1fa5bbd15377fbf26e319cf5189f2080a71ef3
|
| 3 |
+
size 63195406
|
model_fp32.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:35abe1cf4f99a8b9354236d6653d10ac54edae0c05b717acd57fce83a6ac40b1
|
| 3 |
+
size 126333759
|
samples/sample_0.wav
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7ca8e0f9e9d0b0400ce4b7911a20ffa3bb41f42fb18c84fd84c84ff5de50c4cd
|
| 3 |
+
size 798764
|
samples/sample_1.wav
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:df1cea90fac1da28750ac3f06d0448fdb033cc2f2112beb67a75837da3dec6c0
|
| 3 |
+
size 683052
|
samples/sample_2.wav
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d1e77d8790f3ed11b5639f883b816dccf60ef32aa8ceaa9aa596150fe1e21240
|
| 3 |
+
size 759852
|
samples/sample_3.wav
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:172d9f2d583bcd10069efa225a09dae10cd4bc4d6d653a16fb77bcd0f454ec07
|
| 3 |
+
size 832556
|
samples/sample_4.wav
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:eb8fdb2862136fb12a39a4d94ac2c1f15e20f4075f077901061d5029c80d4260
|
| 3 |
+
size 839724
|
symbols.py
ADDED
|
@@ -0,0 +1,57 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
"""
|
| 2 |
+
Defines the set of symbols used in text input to the model.
|
| 3 |
+
Burmese-only vocabulary for StableTTS (character-level, no G2P).
|
| 4 |
+
"""
|
| 5 |
+
|
| 6 |
+
# Padding
|
| 7 |
+
_pad = '_'
|
| 8 |
+
|
| 9 |
+
# ASCII digits (in case clean_text misses some numerals)
|
| 10 |
+
_ascii_digits = '0123456789'
|
| 11 |
+
|
| 12 |
+
# Basic ASCII punctuation
|
| 13 |
+
_ascii_punct = ".,!?'\"-:;()"
|
| 14 |
+
|
| 15 |
+
# Burmese consonants
|
| 16 |
+
_burmese_consonants = (
|
| 17 |
+
'\u1000\u1001\u1002\u1003\u1004\u1005\u1006\u1007\u1008\u1009'
|
| 18 |
+
'\u100A\u100B\u100C\u100D\u100F\u1010\u1011\u1012\u1013\u1014'
|
| 19 |
+
'\u1015\u1016\u1017\u1018\u1019\u101A\u101B\u101C\u101D\u101E'
|
| 20 |
+
'\u101F\u1020\u1021'
|
| 21 |
+
)
|
| 22 |
+
|
| 23 |
+
# Burmese independent vowels
|
| 24 |
+
_burmese_independent = '\u1023\u1024\u1025\u1026\u1027\u1029\u102A'
|
| 25 |
+
|
| 26 |
+
# Burmese dependent vowels & signs
|
| 27 |
+
_burmese_signs = (
|
| 28 |
+
'\u102B\u102C\u102D\u102E\u102F\u1030\u1031\u1032'
|
| 29 |
+
'\u1036\u1037\u1038\u1039\u103A'
|
| 30 |
+
'\u103B\u103C\u103D\u103E\u103F'
|
| 31 |
+
)
|
| 32 |
+
|
| 33 |
+
# Burmese digits
|
| 34 |
+
_burmese_digits = '\u1040\u1041\u1042\u1043\u1044\u1045\u1046\u1047\u1048\u1049'
|
| 35 |
+
|
| 36 |
+
# Burmese punctuation
|
| 37 |
+
_burmese_punct = '\u104A\u104B\u104C\u104D\u104E\u104F'
|
| 38 |
+
|
| 39 |
+
# Extension marker (rare)
|
| 40 |
+
_burmese_ext = '\u1064'
|
| 41 |
+
|
| 42 |
+
# Export all symbols:
|
| 43 |
+
symbols = (
|
| 44 |
+
[_pad]
|
| 45 |
+
+ list(_ascii_digits)
|
| 46 |
+
+ list(_ascii_punct)
|
| 47 |
+
+ list(_burmese_consonants)
|
| 48 |
+
+ list(_burmese_independent)
|
| 49 |
+
+ list(_burmese_signs)
|
| 50 |
+
+ list(_burmese_digits)
|
| 51 |
+
+ list(_burmese_punct)
|
| 52 |
+
+ list(_burmese_ext)
|
| 53 |
+
+ [' ']
|
| 54 |
+
)
|
| 55 |
+
|
| 56 |
+
# Special symbol ids
|
| 57 |
+
SPACE_ID = symbols.index(" ")
|
transcripts.json
ADDED
|
@@ -0,0 +1,194 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"ship_step": 81000,
|
| 3 |
+
"language": "my",
|
| 4 |
+
"sample_rate": 44100,
|
| 5 |
+
"samples": [
|
| 6 |
+
{
|
| 7 |
+
"file": "samples/sample_0.wav",
|
| 8 |
+
"text": "แแผแแบแแฌแแฐแแปแญแฏแธแแปแฌแธแแฌ แกแแฝแแบ แแฅแบแแปแฑแธแแญแแบแแฝแฑแทแแผแฎแธ แงแแบแทแแแบแแปแฌแธแแญแฏ แแปแฐแแปแฐแแพแฌแแพแฌ แแพแญแฏแแบแแพแญแฏแแบแแพแฒแแพแฒแแฒแท แแผแญแฏแแญแฏแแผแแซแแแบ",
|
| 9 |
+
"id": 0
|
| 10 |
+
},
|
| 11 |
+
{
|
| 12 |
+
"file": "samples/sample_1.wav",
|
| 13 |
+
"text": "แแแบแนแแแฌแแซแแพแแบ แแปแฝแแบแแแแฑแฌแท แแผแแบแแฌแแฐแแปแญแฏแธ แแแแบแแญแฏแแบแธแแแบแธแแฐ แกแแปแญแฏแธแแแฎแธแแฑแธ แแ
แบแฅแฎแธ แแผแ
แบแแซแแแบ",
|
| 14 |
+
"id": 1
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"file": "samples/sample_2.wav",
|
| 18 |
+
"text": "แแฎแแฑแท แแปแฝแแบแแแญแฏแทแแฒแท แแฎแแฎแกแแบแ
แบ แ
แแ
แบแแ
แบแแฑแธ แแฑแฌแบแแแบแกแแ
แบแแฑแธแแ
แบแแฏแแญแฏ แกแฑแฌแแบแแผแแบแ
แฝแฌ แแฑแทแแปแแบแท แแแบแแผแฌแธแแญแฏแแบแแฒแทแแซแแแบ",
|
| 19 |
+
"id": 2
|
| 20 |
+
},
|
| 21 |
+
{
|
| 22 |
+
"file": "samples/sample_3.wav",
|
| 23 |
+
"text": "แแฐแแฌแธแแญแฏแแบแธ แแฐแแฌแธแแญแฏแแบแธ แแญแฏแแบแ
แญแแบแแพแ
แบแแฌ แแปแแบแธแแฌแแฝแพแแบแแแบแธแแผแฎแธ แ
แฎแธแแฝแฌแธแแฌแแบแแฌแแแฝแฑ แแฎแแฑแกแแฌแธ แแผแฎแธแแฝแฌแธ แแญแฏแธแแแบแแญแฏแแบแแผแแซแ
แฑ",
|
| 24 |
+
"id": 3
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"file": "samples/sample_4.wav",
|
| 28 |
+
"text": "แแฎแกแแถแแฏแแบแ
แแ
แบแแฑแธแแญแฏ แกแแฏแถแธแแผแฏแแฐแแญแฏแแบแธ แแปแแบแธแแฌแแปแแบแธแแฌแแผแฎแธ แแญแฏแแฌแแแนแแแฝแฑ แแแฏแถแธแแแแแบแธ แแผแแบแทแ
แฏแถแแญแฏแแบแแผแแซแ
แฑ",
|
| 29 |
+
"id": 4
|
| 30 |
+
}
|
| 31 |
+
],
|
| 32 |
+
"checkpoint_comparison": {
|
| 33 |
+
"description": "Same 5 sentences synthesized by checkpoints at different training stages",
|
| 34 |
+
"checkpoints": [
|
| 35 |
+
8000,
|
| 36 |
+
33000,
|
| 37 |
+
44000,
|
| 38 |
+
55000,
|
| 39 |
+
81000
|
| 40 |
+
],
|
| 41 |
+
"files": [
|
| 42 |
+
{
|
| 43 |
+
"file": "checkpoint_comparison/step_8000_sent_0.wav",
|
| 44 |
+
"checkpoint": 8000,
|
| 45 |
+
"text_id": 0,
|
| 46 |
+
"text": "แแผแแบแแฌแแฐแแปแญแฏแธแแปแฌแธแแฌ แกแแฝแแบ แแฅแบแแปแฑแธแแญแแบแแฝแฑแทแแผแฎแธ แงแแบแทแแแบแแปแฌแธแแญแฏ แแปแฐแแปแฐแแพแฌแแพแฌ แแพแญแฏแแบแแพแญแฏแแบแแพแฒแแพแฒแแฒแท แแผแญแฏแแญแฏแแผแแซแแแบ"
|
| 47 |
+
},
|
| 48 |
+
{
|
| 49 |
+
"file": "checkpoint_comparison/step_8000_sent_1.wav",
|
| 50 |
+
"checkpoint": 8000,
|
| 51 |
+
"text_id": 1,
|
| 52 |
+
"text": "แแแบแนแแแฌแแซแแพแแบ แแปแฝแแบแแแแฑแฌแท แแผแแบแแฌแแฐแแปแญแฏแธ แแแแบแแญแฏแแบแธแแแบแธแแฐ แกแแปแญแฏแธแแแฎแธแแฑแธ แแ
แบแฅแฎแธ แแผแ
แบแแซแแแบ"
|
| 53 |
+
},
|
| 54 |
+
{
|
| 55 |
+
"file": "checkpoint_comparison/step_8000_sent_2.wav",
|
| 56 |
+
"checkpoint": 8000,
|
| 57 |
+
"text_id": 2,
|
| 58 |
+
"text": "แแฎแแฑแท แแปแฝแแบแแแญแฏแทแแฒแท แแฎแแฎแกแแบแ
แบ แ
แแ
แบแแ
แบแแฑแธ แแฑแฌแบแแแบแกแแ
แบแแฑแธแแ
แบแแฏแแญแฏ แกแฑแฌแแบแแผแแบแ
แฝแฌ แแฑแทแแปแแบแท แแแบแแผแฌแธแแญแฏแแบแแฒแทแแซแแแบ"
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"file": "checkpoint_comparison/step_8000_sent_3.wav",
|
| 62 |
+
"checkpoint": 8000,
|
| 63 |
+
"text_id": 3,
|
| 64 |
+
"text": "แแฐแแฌแธแแญแฏแแบแธ แแฐแแฌแธแแญแฏแแบแธ แแญแฏแแบแ
แญแแบแแพแ
แบแแฌ แแปแแบแธแแฌแแฝแพแแบแแแบแธแแผแฎแธ แ
แฎแธแแฝแฌแธแแฌแแบแแฌแแแฝแฑ แแฎแแฑแกแแฌแธ แแผแฎแธแแฝแฌแธ แแญแฏแธแแแบแแญแฏแแบแแผแแซแ
แฑ"
|
| 65 |
+
},
|
| 66 |
+
{
|
| 67 |
+
"file": "checkpoint_comparison/step_8000_sent_4.wav",
|
| 68 |
+
"checkpoint": 8000,
|
| 69 |
+
"text_id": 4,
|
| 70 |
+
"text": "แแฎแกแแถแแฏแแบแ
แแ
แบแแฑแธแแญแฏ แกแแฏแถแธแแผแฏแแฐแแญแฏแแบแธ แแปแแบแธแแฌแแปแแบแธแแฌแแผแฎแธ แแญแฏแแฌแแแนแแแฝแฑ แแแฏแถแธแแแแแบแธ แแผแแบแทแ
แฏแถแแญแฏแแบแแผแแซแ
แฑ"
|
| 71 |
+
},
|
| 72 |
+
{
|
| 73 |
+
"file": "checkpoint_comparison/step_33000_sent_0.wav",
|
| 74 |
+
"checkpoint": 33000,
|
| 75 |
+
"text_id": 0,
|
| 76 |
+
"text": "แแผแแบแแฌแแฐแแปแญแฏแธแแปแฌแธแแฌ แกแแฝแแบ แแฅแบแแปแฑแธแแญแแบแแฝแฑแทแแผแฎแธ แงแแบแทแแแบแแปแฌแธแแญแฏ แแปแฐแแปแฐแแพแฌแแพแฌ แแพแญแฏแแบแแพแญแฏแแบแแพแฒแแพแฒแแฒแท แแผแญแฏแแญแฏแแผแแซแแแบ"
|
| 77 |
+
},
|
| 78 |
+
{
|
| 79 |
+
"file": "checkpoint_comparison/step_33000_sent_1.wav",
|
| 80 |
+
"checkpoint": 33000,
|
| 81 |
+
"text_id": 1,
|
| 82 |
+
"text": "แแแบแนแแแฌแแซแแพแแบ แแปแฝแแบแแแแฑแฌแท แแผแแบแแฌแแฐแแปแญแฏแธ แแแแบแแญแฏแแบแธแแแบแธแแฐ แกแแปแญแฏแธแแแฎแธแแฑแธ แแ
แบแฅแฎแธ แแผแ
แบแแซแแแบ"
|
| 83 |
+
},
|
| 84 |
+
{
|
| 85 |
+
"file": "checkpoint_comparison/step_33000_sent_2.wav",
|
| 86 |
+
"checkpoint": 33000,
|
| 87 |
+
"text_id": 2,
|
| 88 |
+
"text": "แแฎแแฑแท แแปแฝแแบแแแญแฏแทแแฒแท แแฎแแฎแกแแบแ
แบ แ
แแ
แบแแ
แบแแฑแธ แแฑแฌแบแแแบแกแแ
แบแแฑแธแแ
แบแแฏแแญแฏ แกแฑแฌแแบแแผแแบแ
แฝแฌ แแฑแทแแปแแบแท แแแบแแผแฌแธแแญแฏแแบแแฒแทแแซแแแบ"
|
| 89 |
+
},
|
| 90 |
+
{
|
| 91 |
+
"file": "checkpoint_comparison/step_33000_sent_3.wav",
|
| 92 |
+
"checkpoint": 33000,
|
| 93 |
+
"text_id": 3,
|
| 94 |
+
"text": "แแฐแแฌแธแแญแฏแแบแธ แแฐแแฌแธแแญแฏแแบแธ แแญแฏแแบแ
แญแแบแแพแ
แบแแฌ แแปแแบแธแแฌแแฝแพแแบแแแบแธแแผแฎแธ แ
แฎแธแแฝแฌแธแแฌแแบแแฌแแแฝแฑ แแฎแแฑแกแแฌแธ แแผแฎแธแแฝแฌแธ แแญแฏแธแแแบแแญแฏแแบแแผแแซแ
แฑ"
|
| 95 |
+
},
|
| 96 |
+
{
|
| 97 |
+
"file": "checkpoint_comparison/step_33000_sent_4.wav",
|
| 98 |
+
"checkpoint": 33000,
|
| 99 |
+
"text_id": 4,
|
| 100 |
+
"text": "แแฎแกแแถแแฏแแบแ
แแ
แบแแฑแธแแญแฏ แกแแฏแถแธแแผแฏแแฐแแญแฏแแบแธ แแปแแบแธแแฌแแปแแบแธแแฌแแผแฎแธ แแญแฏแแฌแแแนแแแฝแฑ แแแฏแถแธแแแแแบแธ แแผแแบแทแ
แฏแถแแญแฏแแบแแผแแซแ
แฑ"
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"file": "checkpoint_comparison/step_44000_sent_0.wav",
|
| 104 |
+
"checkpoint": 44000,
|
| 105 |
+
"text_id": 0,
|
| 106 |
+
"text": "แแผแแบแแฌแแฐแแปแญแฏแธแแปแฌแธแแฌ แกแแฝแแบ แแฅแบแแปแฑแธแแญแแบแแฝแฑแทแแผแฎแธ แงแแบแทแแแบแแปแฌแธแแญแฏ แแปแฐแแปแฐแแพแฌแแพแฌ แแพแญแฏแแบแแพแญแฏแแบแแพแฒแแพแฒแแฒแท แแผแญแฏแแญแฏแแผแแซแแแบ"
|
| 107 |
+
},
|
| 108 |
+
{
|
| 109 |
+
"file": "checkpoint_comparison/step_44000_sent_1.wav",
|
| 110 |
+
"checkpoint": 44000,
|
| 111 |
+
"text_id": 1,
|
| 112 |
+
"text": "แแแบแนแแแฌแแซแแพแแบ แแปแฝแแบแแแแฑแฌแท แแผแแบแแฌแแฐแแปแญแฏแธ แแแแบแแญแฏแแบแธแแแบแธแแฐ แกแแปแญแฏแธแแแฎแธแแฑแธ แแ
แบแฅแฎแธ แแผแ
แบแแซแแแบ"
|
| 113 |
+
},
|
| 114 |
+
{
|
| 115 |
+
"file": "checkpoint_comparison/step_44000_sent_2.wav",
|
| 116 |
+
"checkpoint": 44000,
|
| 117 |
+
"text_id": 2,
|
| 118 |
+
"text": "แแฎแแฑแท แแปแฝแแบแแแญแฏแทแแฒแท แแฎแแฎแกแแบแ
แบ แ
แแ
แบแแ
แบแแฑแธ แแฑแฌแบแแแบแกแแ
แบแแฑแธแแ
แบแแฏแแญแฏ แกแฑแฌแแบแแผแแบแ
แฝแฌ แแฑแทแแปแแบแท แแแบแแผแฌแธแแญแฏแแบแแฒแทแแซแแแบ"
|
| 119 |
+
},
|
| 120 |
+
{
|
| 121 |
+
"file": "checkpoint_comparison/step_44000_sent_3.wav",
|
| 122 |
+
"checkpoint": 44000,
|
| 123 |
+
"text_id": 3,
|
| 124 |
+
"text": "แแฐแแฌแธแแญแฏแแบแธ แแฐแแฌแธแแญแฏแแบแธ แแญแฏแแบแ
แญแแบแแพแ
แบแแฌ แแปแแบแธแแฌแแฝแพแแบแแแบแธแแผแฎแธ แ
แฎแธแแฝแฌแธแแฌแแบแแฌแแแฝแฑ แแฎแแฑแกแแฌแธ แแผแฎแธแแฝแฌแธ แแญแฏแธแแแบแแญแฏแแบแแผแแซแ
แฑ"
|
| 125 |
+
},
|
| 126 |
+
{
|
| 127 |
+
"file": "checkpoint_comparison/step_44000_sent_4.wav",
|
| 128 |
+
"checkpoint": 44000,
|
| 129 |
+
"text_id": 4,
|
| 130 |
+
"text": "แแฎแกแแถแแฏแแบแ
แแ
แบแแฑแธแแญแฏ แกแแฏแถแธแแผแฏแแฐแแญแฏแแบแธ แแปแแบแธแแฌแแปแแบแธแแฌแแผแฎแธ แแญแฏแแฌแแแนแแแฝแฑ แแแฏแถแธแแแแแบแธ แแผแแบแทแ
แฏแถแแญแฏแแบแแผแแซแ
แฑ"
|
| 131 |
+
},
|
| 132 |
+
{
|
| 133 |
+
"file": "checkpoint_comparison/step_55000_sent_0.wav",
|
| 134 |
+
"checkpoint": 55000,
|
| 135 |
+
"text_id": 0,
|
| 136 |
+
"text": "แแผแแบแแฌแแฐแแปแญแฏแธแแปแฌแธแแฌ แกแแฝแแบ แแฅแบแแปแฑแธแแญแแบแแฝแฑแทแแผแฎแธ แงแแบแทแแแบแแปแฌแธแแญแฏ แแปแฐแแปแฐแแพแฌแแพแฌ แแพแญแฏแแบแแพแญแฏแแบแแพแฒแแพแฒแแฒแท แแผแญแฏแแญแฏแแผแแซแแแบ"
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"file": "checkpoint_comparison/step_55000_sent_1.wav",
|
| 140 |
+
"checkpoint": 55000,
|
| 141 |
+
"text_id": 1,
|
| 142 |
+
"text": "แแแบแนแแแฌแแซแแพแแบ แแปแฝแแบแแแแฑแฌแท แแผแแบแแฌแแฐแแปแญแฏแธ แแแแบแแญแฏแแบแธแแแบแธแแฐ แกแแปแญแฏแธแแแฎแธแแฑแธ แแ
แบแฅแฎแธ แแผแ
แบแแซแแแบ"
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"file": "checkpoint_comparison/step_55000_sent_2.wav",
|
| 146 |
+
"checkpoint": 55000,
|
| 147 |
+
"text_id": 2,
|
| 148 |
+
"text": "แแฎแแฑแท แแปแฝแแบแแแญแฏแทแแฒแท แแฎแแฎแกแแบแ
แบ แ
แแ
แบแแ
แบแแฑแธ แแฑแฌแบ๏ฟฝ๏ฟฝ๏ฟฝแแบแกแแ
แบแแฑแธแแ
แบแแฏแแญแฏ แกแฑแฌแแบแแผแแบแ
แฝแฌ แแฑแทแแปแแบแท แแแบแแผแฌแธแแญแฏแแบแแฒแทแแซแแแบ"
|
| 149 |
+
},
|
| 150 |
+
{
|
| 151 |
+
"file": "checkpoint_comparison/step_55000_sent_3.wav",
|
| 152 |
+
"checkpoint": 55000,
|
| 153 |
+
"text_id": 3,
|
| 154 |
+
"text": "แแฐแแฌแธแแญแฏแแบแธ แแฐแแฌแธแแญแฏแแบแธ แแญแฏแแบแ
แญแแบแแพแ
แบแแฌ แแปแแบแธแแฌแแฝแพแแบแแแบแธแแผแฎแธ แ
แฎแธแแฝแฌแธแแฌแแบแแฌแแแฝแฑ แแฎแแฑแกแแฌแธ แแผแฎแธแแฝแฌแธ แแญแฏแธแแแบแแญแฏแแบแแผแแซแ
แฑ"
|
| 155 |
+
},
|
| 156 |
+
{
|
| 157 |
+
"file": "checkpoint_comparison/step_55000_sent_4.wav",
|
| 158 |
+
"checkpoint": 55000,
|
| 159 |
+
"text_id": 4,
|
| 160 |
+
"text": "แแฎแกแแถแแฏแแบแ
แแ
แบแแฑแธแแญแฏ แกแแฏแถแธแแผแฏแแฐแแญแฏแแบแธ แแปแแบแธแแฌแแปแแบแธแแฌแแผแฎแธ แแญแฏแแฌแแแนแแแฝแฑ แแแฏแถแธแแแแแบแธ แแผแแบแทแ
แฏแถแแญแฏแแบแแผแแซแ
แฑ"
|
| 161 |
+
},
|
| 162 |
+
{
|
| 163 |
+
"file": "checkpoint_comparison/step_81000_sent_0.wav",
|
| 164 |
+
"checkpoint": 81000,
|
| 165 |
+
"text_id": 0,
|
| 166 |
+
"text": "แแผแแบแแฌแแฐแแปแญแฏแธแแปแฌแธแแฌ แกแแฝแแบ แแฅแบแแปแฑแธแแญแแบแแฝแฑแทแแผแฎแธ แงแแบแทแแแบแแปแฌแธแแญแฏ แแปแฐแแปแฐแแพแฌแแพแฌ แแพแญแฏแแบแแพแญแฏแแบแแพแฒแแพแฒแแฒแท แแผแญแฏแแญแฏแแผแแซแแแบ"
|
| 167 |
+
},
|
| 168 |
+
{
|
| 169 |
+
"file": "checkpoint_comparison/step_81000_sent_1.wav",
|
| 170 |
+
"checkpoint": 81000,
|
| 171 |
+
"text_id": 1,
|
| 172 |
+
"text": "แแแบแนแแแฌแแซแแพแแบ แแปแฝแแบแแแแฑแฌแท แแผแแบแแฌแแฐแแปแญแฏแธ แแแแบแแญแฏแแบแธแแแบแธแแฐ แกแแปแญแฏแธแแแฎแธแแฑแธ แแ
แบแฅแฎแธ แแผแ
แบแแซแแแบ"
|
| 173 |
+
},
|
| 174 |
+
{
|
| 175 |
+
"file": "checkpoint_comparison/step_81000_sent_2.wav",
|
| 176 |
+
"checkpoint": 81000,
|
| 177 |
+
"text_id": 2,
|
| 178 |
+
"text": "แแฎแแฑแท แแปแฝแแบแแแญแฏแทแแฒแท แแฎแแฎแกแแบแ
แบ แ
แแ
แบแแ
แบแแฑแธ แแฑแฌแบแแแบแกแแ
แบแแฑแธแแ
แบแแฏแแญแฏ แกแฑแฌแแบแแผแแบแ
แฝแฌ แแฑแทแแปแแบแท แแแบแแผแฌแธแแญแฏแแบแแฒแทแแซแแแบ"
|
| 179 |
+
},
|
| 180 |
+
{
|
| 181 |
+
"file": "checkpoint_comparison/step_81000_sent_3.wav",
|
| 182 |
+
"checkpoint": 81000,
|
| 183 |
+
"text_id": 3,
|
| 184 |
+
"text": "แแฐแแฌแธแแญแฏแแบแธ แแฐแแฌแธแแญแฏแแบแธ แแญแฏแแบแ
แญแแบแแพแ
แบแแฌ แแปแแบแธแแฌแแฝแพแแบแแแบแธแแผแฎแธ แ
แฎแธแแฝแฌแธแแฌแแบแแฌแแแฝแฑ แแฎแแฑแกแแฌแธ แแผแฎแธแแฝแฌแธ แแญแฏแธแแแบแแญแฏแแบแแผแแซแ
แฑ"
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"file": "checkpoint_comparison/step_81000_sent_4.wav",
|
| 188 |
+
"checkpoint": 81000,
|
| 189 |
+
"text_id": 4,
|
| 190 |
+
"text": "แแฎแกแแถแแฏแแบแ
แแ
แบแแฑแธแแญแฏ แกแแฏแถแธแแผแฏแแฐแแญแฏแแบแธ แแปแแบแธแแฌแแปแแบแธแแฌแแผแฎแธ แแญแฏแแฌแแแนแแแฝแฑ แแแฏแถแธแแแแแบแธ แแผแแบแทแ
แฏแถแแญแฏแแบแแผแแซแ
แฑ"
|
| 191 |
+
}
|
| 192 |
+
]
|
| 193 |
+
}
|
| 194 |
+
}
|
vocab.txt
ADDED
|
@@ -0,0 +1,98 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
_
|
| 2 |
+
0
|
| 3 |
+
1
|
| 4 |
+
2
|
| 5 |
+
3
|
| 6 |
+
4
|
| 7 |
+
5
|
| 8 |
+
6
|
| 9 |
+
7
|
| 10 |
+
8
|
| 11 |
+
9
|
| 12 |
+
.
|
| 13 |
+
,
|
| 14 |
+
!
|
| 15 |
+
?
|
| 16 |
+
'
|
| 17 |
+
"
|
| 18 |
+
-
|
| 19 |
+
:
|
| 20 |
+
;
|
| 21 |
+
(
|
| 22 |
+
)
|
| 23 |
+
แ
|
| 24 |
+
แ
|
| 25 |
+
แ
|
| 26 |
+
แ
|
| 27 |
+
แ
|
| 28 |
+
แ
|
| 29 |
+
แ
|
| 30 |
+
แ
|
| 31 |
+
แ
|
| 32 |
+
แ
|
| 33 |
+
แ
|
| 34 |
+
แ
|
| 35 |
+
แ
|
| 36 |
+
แ
|
| 37 |
+
แ
|
| 38 |
+
แ
|
| 39 |
+
แ
|
| 40 |
+
แ
|
| 41 |
+
แ
|
| 42 |
+
แ
|
| 43 |
+
แ
|
| 44 |
+
แ
|
| 45 |
+
แ
|
| 46 |
+
แ
|
| 47 |
+
แ
|
| 48 |
+
แ
|
| 49 |
+
แ
|
| 50 |
+
แ
|
| 51 |
+
แ
|
| 52 |
+
แ
|
| 53 |
+
แ
|
| 54 |
+
แ
|
| 55 |
+
แก
|
| 56 |
+
แฃ
|
| 57 |
+
แค
|
| 58 |
+
แฅ
|
| 59 |
+
แฆ
|
| 60 |
+
แง
|
| 61 |
+
แฉ
|
| 62 |
+
แช
|
| 63 |
+
แซ
|
| 64 |
+
แฌ
|
| 65 |
+
แญ
|
| 66 |
+
แฎ
|
| 67 |
+
แฏ
|
| 68 |
+
แฐ
|
| 69 |
+
แฑ
|
| 70 |
+
แฒ
|
| 71 |
+
แถ
|
| 72 |
+
แท
|
| 73 |
+
แธ
|
| 74 |
+
แน
|
| 75 |
+
แบ
|
| 76 |
+
แป
|
| 77 |
+
แผ
|
| 78 |
+
แฝ
|
| 79 |
+
แพ
|
| 80 |
+
แฟ
|
| 81 |
+
แ
|
| 82 |
+
แ
|
| 83 |
+
แ
|
| 84 |
+
แ
|
| 85 |
+
แ
|
| 86 |
+
แ
|
| 87 |
+
แ
|
| 88 |
+
แ
|
| 89 |
+
แ
|
| 90 |
+
แ
|
| 91 |
+
แ
|
| 92 |
+
แ
|
| 93 |
+
แ
|
| 94 |
+
แ
|
| 95 |
+
แ
|
| 96 |
+
แ
|
| 97 |
+
แค
|
| 98 |
+
<space>
|
vocos.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0cfbc25b8ac9e2994867af215b4cb5577cf60a52556c3bd1156b6954d0e9db98
|
| 3 |
+
size 56666508
|