Spaces:
Build error
Build error
File size: 3,166 Bytes
16c455e c7e9ab6 44ea321 c7e9ab6 23eacd8 c7e9ab6 ec1fd4a c7e9ab6 ec1fd4a c7e9ab6 ec1fd4a c7e9ab6 ec1fd4a c7e9ab6 ec1fd4a c7e9ab6 55633f7 04bb41f f982d07 c7e9ab6 |
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 |
import os
os.system("pip install gradio==2.8.0b2")
import gradio as gr
import librosa
from transformers import AutoFeatureExtractor, AutoTokenizer, SpeechEncoderDecoderModel
model_name = "facebook/wav2vec2-xls-r-2b-en-to-15"
feature_extractor = AutoFeatureExtractor.from_pretrained(model_name)
tokenizer = AutoTokenizer.from_pretrained(model_name, use_fast=False)
model = SpeechEncoderDecoderModel.from_pretrained(model_name)
def process_audio_file(file):
data, sr = librosa.load(file)
if sr != 16000:
data = librosa.resample(data, sr, 16000)
print(data.shape)
input_values = feature_extractor(data, return_tensors="pt").input_values
return input_values
def transcribe(file_mic, file_upload, target_language):
target_code = target_language.split("(")[-1].split(")")[0]
forced_bos_token_id = MAPPING[target_code]
warn_output = ""
if (file_mic is not None) and (file_upload is not None):
warn_output = "WARNING: You've uploaded an audio file and used the microphone. The recorded file from the microphone will be used and the uploaded audio will be discarded.\n"
file = file_mic
elif (file_mic is None) and (file_upload is None):
return "ERROR: You have to either use the microphone or upload an audio file"
elif file_mic is not None:
file = file_mic
else:
file = file_upload
input_values = process_audio_file(file)
sequences = model.generate(input_values, forced_bos_token_id=forced_bos_token_id, num_beams=1, max_length=30)
transcription = tokenizer.batch_decode(sequences, skip_special_tokens=True)
return warn_output + transcription[0]
target_language = [
"English (en)",
"German (de)",
"Turkish (tr)",
"Persian (fa)",
"Swedish (sv)",
"Mongolian (mn)",
"Chinese (zh)",
"Welsh (cy)",
"Catalan (ca)",
"Slovenian (sl)",
"Estonian (et)",
"Indonesian (id)",
"Arabic (ar)",
"Tamil (ta)",
"Latvian (lv)",
"Japanese (ja)",
]
MAPPING = {
"en": 250004,
"de": 250003,
"tr": 250023,
"fa": 250029,
"sv": 250042,
"mn": 250037,
"zh": 250025,
"cy": 250007,
"ca": 250005,
"sl": 250052,
"et": 250006,
"id": 250032,
"ar": 250001,
"ta": 250044,
"lv": 250017,
"ja": 250012,
}
iface = gr.Interface(
fn=transcribe,
inputs=[
gr.inputs.Audio(source="microphone", type='filepath', optional=True),
gr.inputs.Audio(source="upload", type='filepath', optional=True),
gr.inputs.Dropdown(target_language),
],
outputs="text",
layout="horizontal",
theme="huggingface",
title="XLS-R 2B EN-to-15 Speech Translation",
description="A simple interface to translate from spoken English to 15 written languages.",
article = "<p style='text-align: center'><a href='https://huggingface.co/facebook/wav2vec2-xls-r-2b-en-to-15' target='_blank'>Click to learn more about XLS-R-2B-EN-15 </a> | <a href='https://arxiv.org/abs/2111.09296' target='_blank'> With 🎙️ from Facebook XLS-R </a></p>",
enable_queue=True,
allow_flagging=False,
)
iface.launch()
|