|
from tts_infer.tts import TextToMel, MelToWav |
|
from tts_infer.transliterate import XlitEngine |
|
from tts_infer.num_to_word_on_sent import normalize_nums |
|
|
|
import re |
|
import numpy as np |
|
from scipy.io.wavfile import write |
|
|
|
from mosestokenizer import * |
|
from indicnlp.tokenize import sentence_tokenize |
|
import gradio as gr |
|
|
|
|
|
INDIC = ["as", "bn", "gu", "hi", "kn", "ml", "mr", "or", "pa", "ta", "te"] |
|
|
|
def split_sentences(paragraph, language): |
|
if language == "en": |
|
with MosesSentenceSplitter(language) as splitter: |
|
return splitter([paragraph]) |
|
elif language in INDIC: |
|
return sentence_tokenize.sentence_split(paragraph, lang=language) |
|
|
|
|
|
device='cpu' |
|
text_to_mel = TextToMel(glow_model_dir='vakyansh-tts/tts_infer/odia/glow', device=device) |
|
mel_to_wav = MelToWav(hifi_model_dir='vakyansh-tts/tts_infer/odia/hifi', device=device) |
|
|
|
|
|
def run_tts(text, lang): |
|
final_text = text |
|
mel = text_to_mel.generate_mel(final_text) |
|
audio, sr = mel_to_wav.generate_wav(mel) |
|
write(filename='temp.wav', rate=sr, data=audio) |
|
return (sr, audio) |
|
|
|
def run_tts_paragraph(text, lang): |
|
audio_list = [] |
|
split_sentences_list = split_sentences(text, language='hi') |
|
|
|
for sent in split_sentences_list: |
|
sr, audio = run_tts(sent, lang) |
|
audio_list.append(audio) |
|
|
|
concatenated_audio = np.concatenate([i for i in audio_list]) |
|
write(filename='temp_long.wav', rate=sr, data=concatenated_audio) |
|
return (sr, concatenated_audio) |
|
|
|
|
|
|
|
_, audio = run_tts("ଆମେ ଦୁଖିତ, ଆପଣଙ୍କର ଚିନ୍ତାଧାରାକୁ ସମାଧାନ କରିବାରେ ଅସମର୍ଥ, ଆମେ ଆପଣଙ୍କ ସହ ଯୋଗାଯୋଗ କରିବାକୁ ୱାର୍କସପ୍ଦ ଦଳକୁ କହିବୁ, ତୁମର ଦିନ ଶୁଭମୟ ହଉ.", "or") |
|
|
|
|
|
options = ["Odia"] |
|
|
|
newOptions = ["Male","Female"] |
|
language = gr.Dropdown(options,label="Select language") |
|
gender = gr.Dropdown(newOptions,label="Select Voice") |
|
input = gr.Textbox( |
|
label="Input from model will appear here:", |
|
lines=5 |
|
) |
|
output = gr.Audio(label="Output from model will appear here:", type="filepath") |
|
|
|
gr.Interface(run_tts, inputs = [input,language], outputs=output, |
|
streaming=True, interactive=True, |
|
analytics_enabled=False, show_tips=False, enable_queue=True).launch(inline=False); |
