rutsam commited on
Commit
9506ac8
1 Parent(s): c7f1d42

use soundfile instead of librosa to write files

Browse files
Files changed (2) hide show
  1. app.py +0 -5
  2. nemo_asr.py +2 -1
app.py CHANGED
@@ -8,11 +8,6 @@ from transformers import Wav2Vec2ProcessorWithLM, Wav2Vec2CTCTokenizer, Wav2Vec2
8
 
9
  from engine import SpeechToTextEngine
10
  import wave
11
- import gradio as gr
12
- import librosa
13
- import soundfile as sf
14
- import warnings
15
-
16
  from nemo_asr import transcribe
17
 
18
 
 
8
 
9
  from engine import SpeechToTextEngine
10
  import wave
 
 
 
 
 
11
  from nemo_asr import transcribe
12
 
13
 
nemo_asr.py CHANGED
@@ -2,6 +2,7 @@ import aiofiles
2
  import nemo
3
  import nemo.collections.asr as nemo_asr
4
  import librosa
 
5
 
6
 
7
  def transcribe(file, modelName="stt_rw_conformer_transducer_large"):
@@ -23,7 +24,7 @@ def transcribe(file, modelName="stt_rw_conformer_transducer_large"):
23
  # print(f"Audio in {fname} was recognized as: {transcription}")
24
  # print(transcription[0])
25
  x, sr = librosa.load(file.name)
26
- librosa.output.write_wav("Test.wav", x, sr=16000, norm=False)
27
  transcription= asr_model.transcribe(["Test.wav"])
28
  print(transcription)
29
  return {"text": transcription, "filename": file.name}
 
2
  import nemo
3
  import nemo.collections.asr as nemo_asr
4
  import librosa
5
+ import soundfile as sf
6
 
7
 
8
  def transcribe(file, modelName="stt_rw_conformer_transducer_large"):
 
24
  # print(f"Audio in {fname} was recognized as: {transcription}")
25
  # print(transcription[0])
26
  x, sr = librosa.load(file.name)
27
+ sf.write("Test.wav", x, sr=16000, norm=False)
28
  transcription= asr_model.transcribe(["Test.wav"])
29
  print(transcription)
30
  return {"text": transcription, "filename": file.name}