File size: 1,498 Bytes
c106aba
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
06628a1
 
c106aba
 
06628a1
c106aba
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
06628a1
 
 
c106aba
 
 
06628a1
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
import soundfile as sf
import torch
from transformers import Wav2Vec2ForCTC, Wav2Vec2Processor
import argparse
from glob import glob
import subprocess
import gradio as gr


def get_filename(wav_file):
    filename_local = wav_file.split('/')[-1][:-4]
    filename_new = '/tmp/'+filename_local+'_16.wav'
    
    
    subprocess.call(["sox {} -r {} -b 16 -c 1 {}".format(wav_file, str(16000), filename_new)], shell=True)
    return filename_new



def parse_transcription(wav_file):
    # load pretrained model
    
    # load audio

    
    wav_file = get_filename(wav_file.name)
    audio_input, sample_rate = sf.read(wav_file)
    #test_file = resampler(test_file[0])

    # pad input values and return pt tensor
    input_values = processor(audio_input, sampling_rate=16_000, return_tensors="pt").input_values

    # INFERENCE
    # retrieve logits & take argmax
    logits = model(input_values).logits
    predicted_ids = torch.argmax(logits, dim=-1)

    # transcribe
    transcription = processor.decode(predicted_ids[0], skip_special_tokens=True)
    return transcription
    

processor = Wav2Vec2Processor.from_pretrained("Harveenchadha/vakyansh-wav2vec2-hindi-him-4200")
model = Wav2Vec2ForCTC.from_pretrained("Harveenchadha/vakyansh-wav2vec2-hindi-him-4200")
    
input = gr.inputs.Audio(source="microphone", type="file") 
gr.Interface(parse_transcription, inputs = input,  outputs="text", 
             analytics_enabled=False, show_tips=False, enable_queue=True).launch(inline=False);