Spaces:
Runtime error
Runtime error
ktangri
commited on
Commit
·
a263f35
1
Parent(s):
a893c73
Add speech transcription
Browse files
app.py
ADDED
|
@@ -0,0 +1,28 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import gradio as gr
|
| 2 |
+
from transformers import pipeline
|
| 3 |
+
from librosa import load, resample
|
| 4 |
+
|
| 5 |
+
|
| 6 |
+
asr_model = 'facebook/wav2vec2-base-960h'
|
| 7 |
+
|
| 8 |
+
asr = pipeline('automatic-speech-recognition', model=asr_model, feature_extractor=asr_model)
|
| 9 |
+
|
| 10 |
+
def transcribe(filepath):
|
| 11 |
+
speech, sampling_rate = load(filepath)
|
| 12 |
+
if sampling_rate != 16000:
|
| 13 |
+
speech = resample(speech, sampling_rate, 16000)
|
| 14 |
+
text = asr(speech)['text']
|
| 15 |
+
return text
|
| 16 |
+
|
| 17 |
+
mic = gr.inputs.Audio(source='microphone', type='filepath', label='Speech input', optional=False)
|
| 18 |
+
|
| 19 |
+
transcript = gr.outputs.Textbox(type='auto', label='Transcription')
|
| 20 |
+
|
| 21 |
+
iface = gr.Interface(
|
| 22 |
+
theme='huggingface',
|
| 23 |
+
description='Testing transcription',
|
| 24 |
+
fn=transcribe,
|
| 25 |
+
inputs=[mic],
|
| 26 |
+
outputs=[transcript]
|
| 27 |
+
)
|
| 28 |
+
iface.launch()
|