import os os.system("pip uninstall -y gradio") os.system("pip install gradio==2.7") import gradio as gr #os.system('git clone https://github.com/Open-Speech-EkStep/vakyansh-tts') os.chdir('vakyansh_tts') os.system('bash install.sh') os.system('python setup.py bdist_wheel') os.system('pip install -e .') os.chdir('tts_infer') os.system('mkdir translit_models') os.chdir('translit_models') os.system('wget -q https://storage.googleapis.com/vakyaansh-open-models/translit_models/default_lineup.json') os.system('mkdir hindi') os.chdir('hindi') os.system('wget -q https://storage.googleapis.com/vakyaansh-open-models/translit_models/hindi/hindi_transliteration.zip') os.system('unzip hindi_transliteration') os.system('wget -q https://storage.googleapis.com/vakyansh-open-models/tts/hindi/hi-IN/female_voice_0/glow.zip') os.system('unzip glow.zip') os.system('wget -q https://storage.googleapis.com/vakyansh-open-models/tts/hindi/hi-IN/female_voice_0/hifi.zip') os.system('unzip hifi.zip') os.system('rm glow.zip') os.system('rm hifi.zip') os.system('mkdir male') os.chdir('male') os.system('wget -q https://storage.googleapis.com/vakyansh-open-models/tts/hindi/hi-IN/male_voice_1/glow.zip') os.system('unzip glow.zip') os.system('wget -q https://storage.googleapis.com/vakyansh-open-models/tts/hindi/hi-IN/male_voice_1/hifi.zip') os.system('unzip hifi.zip') os.system('pwd') os.system('rm glow.zip') os.system('rm hifi.zip') os.system('pip uninstall -y numpy') os.system('pip install numpy==1.19.5') os.system('pip uninstall -y numba') os.system('pip install numba==0.53') os.chdir('/home/user/app/') os.system('pwd') #print('hello') from vakyansh_tts.tts_infer.tts import TextToMel, MelToWav from vakyansh_tts.tts_infer.transliterate import XlitEngine from vakyansh_tts.tts_infer.num_to_word_on_sent import normalize_nums import re from scipy.io.wavfile import write device = 'cpu' text_to_mel_f = TextToMel(glow_model_dir='/home/user/app/vakyansh_tts/tts_infer/translit_models/hindi/glow_ckp', device=device) mel_to_wav_f = MelToWav(hifi_model_dir='/home/user/app/vakyansh_tts/tts_infer/translit_models/hindi/hifi_ckp', device=device) text_to_mel_m = TextToMel(glow_model_dir='/home/user/app/vakyansh_tts/tts_infer/translit_models/hindi/male/glow_ckp', device=device) mel_to_wav_m = MelToWav(hifi_model_dir='/home/user/app/vakyansh_tts/tts_infer/translit_models/hindi/male/hifi_ckp', device=device) def translit(text, lang): reg = re.compile(r'[a-zA-Z]') engine = XlitEngine(lang) words = [engine.translit_word(word, topk=1)[lang][0] if reg.match(word) else word for word in text.split()] updated_sent = ' '.join(words) return updated_sent def run_tts(text, gender): print("Original Text from user: ", text) lang='hi' text = text.replace('।', '.') # only for hindi models text_num_to_word = normalize_nums(text, lang) # converting numbers to words in lang text_num_to_word_and_transliterated = translit(text_num_to_word, lang) # transliterating english words to lang print("Text after preprocessing: ", text_num_to_word_and_transliterated) if gender == 'female': mel = text_to_mel_f.generate_mel(text_num_to_word_and_transliterated) audio, sr = mel_to_wav_f.generate_wav(mel) else: mel = text_to_mel_m.generate_mel(text_num_to_word_and_transliterated) audio, sr = mel_to_wav_m.generate_wav(mel) #write(filename='temp.wav', rate=sr, data=audio) # for saving wav file, if needed return (sr, audio) #_, audio = run_tts('hello my name is harveen') textbox = gr.inputs.Textbox( placeholder="Enter Hindi text here", default="", label="TTS" ) choices = ['male', 'female'] radioBtns = gr.inputs.Radio(choices, type="value", default='male', label=None) op = gr.outputs.Audio(type="numpy", label=None) examples = [['क्रिप्टो करेंसी दरअसल, वित्तीय लेन-देन का एक जरिया है। बिल्कुल भारतीय रुपये और अमेरिकी डॉलर के समान, अंतर सिर्फ इतना है कि यह आभाषी है और दिखाई नहीं देती, न ही आप इसे छू सकते हैं।', 'male'], ['mujhe abhi bhi yakeen nai aa raha ki yeh aise bhi chal sakta hai', 'male'], ['मुझे 26 रुपए दे दो, फिर मेरे पास 50 रुपए हो जाएंगे', 'male']] iface = gr.Interface(fn=run_tts, examples=examples, inputs=[textbox,radioBtns], outputs=op, title='Vakyansh Text To Speech (TTS): Hindi Demo', description = 'Glow TTS + hifi gan. Training Code: https://github.com/Open-Speech-EkStep/vakyansh-tts ' , article = '
Note: This space is running on CPU, inference times will be higher. Please report issues to @harveenchadha twitter. ') iface.launch(enable_queue=True, cache_examples=True)