File size: 1,800 Bytes
3071eeb
 
 
 
 
631e9fa
3071eeb
 
 
 
 
 
631e9fa
 
3071eeb
 
 
631e9fa
3071eeb
 
 
 
631e9fa
 
 
 
 
 
3071eeb
 
 
 
 
 
 
 
 
 
 
631e9fa
3071eeb
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
import gradio as gr
import numpy as np
import torch
from datasets import load_dataset

from transformers import VitsModel, VitsTokenizer, pipeline


device = "cuda:0" if torch.cuda.is_available() else "cpu"

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": "nl"})
    return outputs["text"]


def synthesise(text):
    inputs = tokenizer(text, return_tensors="pt")
    with torch.no_grad():
        outputs = model(inputs["input_ids"])

    speech = outputs.audio[0]
    return speech


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 Italian to Dutch speech translation using OpenAI Whisper and MMS models
"""

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()