File size: 1,878 Bytes
0b452e3
484ff18
0b452e3
 
e3a183b
 
b2a5d55
 
da142ce
 
 
 
 
 
 
 
 
 
 
 
 
 
 
d77f418
 
0b452e3
 
 
 
 
 
 
6596104
0b452e3
 
 
b2a5d55
 
 
da142ce
0b452e3
b867ef9
 
 
 
0b452e3
 
6596104
 
0b452e3
6596104
b867ef9
0b452e3
 
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
import gradio as gr
from transformers import pipeline
import torch
import librosa
import json

max_duration = int(30 * sr)

def load_model(model_name = "cawoylel/windanam_mms-1b-tts_v2"):
  """
  Function to load model from hugging face
  """
  pipe = pipeline("automatic-speech-recognition", model=model_name)
  return pipe

pipeline = load_model()

def transcribe_audio(sample):
  """
  Transcribe audio
  """
  transcription = pipeline(sample)
  return transcription["text"]

def transcribe(audio_file_mic=None, audio_file_upload=None):
    if audio_file_mic:
        audio_file = audio_file_mic
    elif audio_file_upload:
        audio_file = audio_file_upload
    else:
        return "Please upload an audio file or record one"

    # Make sure audio is 16kHz
    speech, sample_rate = librosa.load(audio_file)
    if sample_rate != 16000:
        speech = librosa.resample(speech, orig_sr=sample_rate, target_sr=16000)
    duration = librosa.get_duration(speech)
    if duration > 30:
        speech = speech[:max_duration]
    return transcribe_audio(speech)


description = '''Automatic Speech Recognition with [MMS](https://ai.facebook.com/blog/multilingual-model-speech-recognition/) (Massively Multilingual Speech) by Meta.
Supports [1162 languages](https://dl.fbaipublicfiles.com/mms/misc/language_coverage_mms.html). Read the paper for more details: [Scaling Speech Technology to 1,000+ Languages](https://arxiv.org/abs/2305.13516).'''

iface = gr.Interface(fn=transcribe,
                     inputs=[
                         gr.Audio(source="microphone", type="filepath", label="Record Audio"),
                         gr.Audio(source="upload", type="filepath", label="Upload Audio"),
                         ],
                     outputs=gr.Textbox(label="Transcription"),
                     description=description
                     )
iface.launch()