File size: 4,814 Bytes
d347764 5cdac43 d347764 5cdac43 d347764 5cdac43 d347764 5cdac43 d347764 5cdac43 d347764 5cdac43 d347764 5cdac43 d347764 f805e49 c6f1d54 f805e49 c737803 d347764 226ec3a d347764 f805e49 d347764 c737803 3946ba6 c737803 5cdac43 |
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 |
import gradio as gr
import numpy as np
import torch
from transformers import (
VitsModel,
VitsTokenizer,
pipeline
)
device = "cuda:0" if torch.cuda.is_available() else "cpu"
# load speech translation checkpoint
asr_pipe = pipeline(
"automatic-speech-recognition",
model="openai/whisper-base",
device=device
)
model = VitsModel.from_pretrained("Matthijs/mms-tts-deu")
tokenizer = VitsTokenizer.from_pretrained("Matthijs/mms-tts-deu")
def translate(audio):
outputs = asr_pipe(
audio,
max_new_tokens=256,
generate_kwargs={"task": "transcribe", "language": "de"}
)
return outputs["text"]
def synthesise(text):
if len(text.strip()) == 0:
return (16000, np.zeros(0).astype(np.int16))
inputs = tokenizer(text, return_tensors="pt")
input_ids = inputs["input_ids"]
with torch.no_grad():
outputs = model(input_ids)
speech = outputs.audio[0]
return speech.cpu()
def speech_to_speech_translation(audio):
translated_text = translate(audio)
synthesised_speech = synthesise(translated_text)
synthesised_speech = (synthesised_speech.numpy() * 32767).astype(np.int16)
return 16000, synthesised_speech
title = "Cascaded STST"
description = """
Demo for cascaded speech-to-speech translation (STST), mapping from source speech in any language to target speech in English. Demo uses OpenAI's [Whisper Base](https://huggingface.co/openai/whisper-base) model for speech translation, and Microsoft's
[SpeechT5 TTS](https://huggingface.co/microsoft/speecht5_tts) model for text-to-speech:
"""
demo = gr.Blocks()
mic_translate = gr.Interface(
fn=speech_to_speech_translation,
inputs=gr.Audio(source="microphone", type="filepath"),
outputs=gr.Audio(label="Generated Speech", type="numpy"),
title=title,
description=description,
)
file_translate = gr.Interface(
fn=speech_to_speech_translation,
inputs=gr.Audio(source="upload", type="filepath"),
outputs=gr.Audio(label="Generated Speech", type="numpy"),
examples=[["./example.wav"]],
title=title,
description=description,
)
with demo:
gr.TabbedInterface([mic_translate, file_translate], ["Microphone", "Audio File"])
demo.launch()import gradio as gr
import numpy as np
import torch
from transformers import (
VitsModel,
VitsTokenizer,
pipeline
)
device = "cuda:0" if torch.cuda.is_available() else "cpu"
# load speech translation checkpoint
asr_pipe = pipeline(
"automatic-speech-recognition",
model="openai/whisper-base",
device=device
)
model = VitsModel.from_pretrained("Matthijs/mms-tts-deu")
tokenizer = VitsTokenizer.from_pretrained("Matthijs/mms-tts-deu")
def translate(audio):
outputs = asr_pipe(
audio,
max_new_tokens=256,
generate_kwargs={"task": "transcribe", "language": "de"}
)
return outputs["text"]
def synthesise(text):
if len(text.strip()) == 0:
return (16000, np.zeros(0).astype(np.int16))
inputs = tokenizer(text, return_tensors="pt")
input_ids = inputs["input_ids"]
with torch.no_grad():
outputs = model(input_ids)
speech = outputs.audio[0]
return speech.cpu()
def speech_to_speech_translation(audio):
translated_text = translate(audio)
synthesised_speech = synthesise(translated_text)
synthesised_speech = (synthesised_speech.numpy() * 32767).astype(np.int16)
return 16000, synthesised_speech
title = "Cascaded STST"
description = """
Demo for cascaded speech-to-speech translation (STST), mapping from source speech in any language to target speech in English. Demo uses OpenAI's [Whisper Base](https://huggingface.co/openai/whisper-base) model for speech translation, and Microsoft's
[SpeechT5 TTS](https://huggingface.co/microsoft/speecht5_tts) model for text-to-speech:
"""
demo = gr.Blocks()
mic_translate = gr.Interface(
fn=speech_to_speech_translation,
inputs=gr.Audio(source="microphone", type="filepath"),
outputs=gr.Audio(label="Generated Speech", type="numpy"),
title=title,
description=description,
)
file_translate = gr.Interface(
fn=speech_to_speech_translation,
inputs=gr.Audio(source="upload", type="filepath"),
outputs=gr.Audio(label="Generated Speech", type="numpy"),
examples=[["./example.wav"]],
title=title,
description=description,
)
with demo:
gr.TabbedInterface([mic_translate, file_translate], ["Microphone", "Audio File"])
demo.launch() |