Spaces:
Running
Running
Add application file
Browse files- app.py +28 -0
- requirements.txt +3 -0
app.py
ADDED
|
@@ -0,0 +1,28 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import gradio as gr
|
| 2 |
+
from transformers import pipeline
|
| 3 |
+
from librosa import load
|
| 4 |
+
|
| 5 |
+
def transcribe(input_audio):
|
| 6 |
+
speech, _ = load(input_audio, sr=16000, mono=True)
|
| 7 |
+
output = pipe(speech, chunk_length_s=30, stride_length_s=5)['text']
|
| 8 |
+
return output
|
| 9 |
+
|
| 10 |
+
pipe = pipeline(
|
| 11 |
+
"automatic-speech-recognition",
|
| 12 |
+
model="GetmanY1/wav2vec2-large-sami-cont-pt-22k-finetuned",
|
| 13 |
+
device="cpu"
|
| 14 |
+
)
|
| 15 |
+
|
| 16 |
+
gradio_app = gr.Interface(
|
| 17 |
+
transcribe,
|
| 18 |
+
gr.Audio(sources=["upload","microphone"]),
|
| 19 |
+
"text",
|
| 20 |
+
)
|
| 21 |
+
|
| 22 |
+
if __name__ == "__main__":
|
| 23 |
+
gradio_app.launch()
|
| 24 |
+
|
| 25 |
+
|
| 26 |
+
|
| 27 |
+
# if __name__ == "__main__":
|
| 28 |
+
# gradio_app.launch()
|
requirements.txt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
transformers
|
| 2 |
+
torch
|
| 3 |
+
librosa
|