Spaces:
Build error
Build error
File size: 5,093 Bytes
f5460b4 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 |
"""
TODO:
+ [x] Load Configuration
+ [ ] Checking
+ [ ] Better saving directory
"""
import numpy as np
from pathlib import Path
import jiwer
import pdb
import torch.nn as nn
import torch
import torchaudio
from transformers import pipeline
# from time import process_time, time
from pathlib import Path
import time
# local import
import sys
from espnet2.bin.tts_inference import Text2Speech
# pdb.set_trace()
device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu")
sys.path.append("src")
import gradio as gr
# ASR part
audio_files = [
str(x)
for x in sorted(
Path(
"/home/kevingeng/Disk2/laronix/laronix_automos/data/20230103_video"
).glob("**/*wav")
)
]
# audio_files = [str(x) for x in sorted(Path("./data/Patient_sil_trim_16k_normed_5_snr_40/Rainbow").glob("**/*wav"))]
transcriber = pipeline(
"automatic-speech-recognition",
model="KevinGeng/PAL_John_128_train_dev_test_seed_1",
)
old_transcriber = pipeline(
"automatic-speech-recognition", "facebook/wav2vec2-base-960h"
)
# transcriber = pipeline("automatic-speech-recognition", model="KevinGeng/PAL_John_128_p326_300_train_dev_test_seed_1")
# 【Female】kan-bayashi ljspeech parallel wavegan
# tts_model = Text2Speech.from_pretrained("espnet/kan-bayashi_ljspeech_vits")
# 【Male】fastspeech2-en-200_speaker-cv4, hifigan vocoder
# pdb.set_trace()
from fairseq.checkpoint_utils import load_model_ensemble_and_task_from_hf_hub
from fairseq.models.text_to_speech.hub_interface import TTSHubInterface
# @title English multi-speaker pretrained model { run: "auto" }
lang = "English"
tag = "kan-bayashi/libritts_xvector_vits"
# vits needs no
vocoder_tag = "parallel_wavegan/vctk_parallel_wavegan.v1.long" # @param ["none", "parallel_wavegan/vctk_parallel_wavegan.v1.long", "parallel_wavegan/vctk_multi_band_melgan.v2", "parallel_wavegan/vctk_style_melgan.v1", "parallel_wavegan/vctk_hifigan.v1", "parallel_wavegan/libritts_parallel_wavegan.v1.long", "parallel_wavegan/libritts_multi_band_melgan.v2", "parallel_wavegan/libritts_hifigan.v1", "parallel_wavegan/libritts_style_melgan.v1"] {type:"string"}
from espnet2.bin.tts_inference import Text2Speech
from espnet2.utils.types import str_or_none
text2speech = Text2Speech.from_pretrained(
model_tag=str_or_none(tag),
vocoder_tag=str_or_none(vocoder_tag),
device="cuda",
use_att_constraint=False,
backward_window=1,
forward_window=3,
speed_control_alpha=1.0,
)
import glob
import os
import numpy as np
import kaldiio
# Get model directory path
from espnet_model_zoo.downloader import ModelDownloader
d = ModelDownloader()
model_dir = os.path.dirname(d.download_and_unpack(tag)["train_config"])
# Speaker x-vector selection
xvector_ark = [
p
for p in glob.glob(
f"{model_dir}/../../dump/**/spk_xvector.ark", recursive=True
)
if "tr" in p
][0]
xvectors = {k: v for k, v in kaldiio.load_ark(xvector_ark)}
spks = list(xvectors.keys())
male_spks = {
"M1": "2300_131720",
"M2": "1320_122612",
"M3": "1188_133604",
"M4": "61_70970",
}
female_spks = {"F1": "2961_961", "F2": "8463_287645", "F3": "121_121726"}
spks = dict(male_spks, **female_spks)
spk_names = sorted(spks.keys())
## 20230224 Mousa: No reference,
def ASRold(audio_file):
reg_text = old_transcriber(audio_file)["text"]
return reg_text
def ASRnew(audio_file, state=""):
# pdb.set_trace()
time.sleep(2)
reg_text = transcriber(audio_file)["text"]
state += reg_text + "\n"
return state, state
def VAD(audio_file):
# pdb.set_trace()
reg_text = transcriber(audio_file)["text"]
return 1
reference_textbox = gr.Textbox(
value="",
placeholder="Input reference here",
label="Reference",
)
recognization_textbox = gr.Textbox(
value="",
placeholder="Output recognization here",
label="recognization_textbox",
)
speaker_option = gr.Radio(choices=spk_names, label="Speaker")
input_audio = gr.Audio(
source="upload", type="filepath", label="Audio_to_Evaluate"
)
output_audio = gr.Audio(
source="upload", file="filepath", label="Synthesized Audio"
)
examples = [
["./samples/001.wav", "M1", ""],
["./samples/002.wav", "M2", ""],
["./samples/003.wav", "F1", ""],
["./samples/004.wav", "F2", ""],
]
def change_audiobox(choice):
if choice == "upload":
input_audio = gr.Audio.update(source="upload", visible=True)
elif choice == "microphone":
input_audio = gr.Audio.update(source="microphone", visible=True)
else:
input_audio = gr.Audio.update(visible=False)
return input_audio
demo = gr.Interface(
fn=ASRnew,
inputs=[
gr.Audio(source="microphone", type="filepath", streaming=True),
"state"
],
outputs=[
"textbox",
"state"
],
live=True)
# ASRnew(["/home/kevingeng/Disk2/laronix/Laronix_ASR_TTS_VC/wav/20221228_video_good_normed_5/take1_001_norm.wav", "state"])
# VAD("/home/kevingeng/Disk2/laronix/Laronix_ASR_TTS_VC/wav/20221228_video_good_normed_5/take1_001_norm.wav")
demo.launch(share=False) |