Spaces:
Running
Running
Update app.py
Browse files
app.py
CHANGED
@@ -1,41 +1,46 @@
|
|
1 |
-
# from speechbrain.inference.ASR import EncoderASR
|
2 |
-
# import gradio as gr
|
3 |
|
4 |
-
# model = EncoderASR.from_hparams("speechbrain/asr-wav2vec2-dvoice-wolof")
|
5 |
|
6 |
-
# def transcribe(audio):
|
7 |
-
# return model.transcribe_file(audio.name)
|
8 |
|
9 |
|
10 |
-
# demo = gr.Interface(fn=transcribe, inputs="file", outputs="text",
|
11 |
-
# title="Transcription automatique du wolof",
|
12 |
-
# description="Ce modèle transcrit un fichier audio en wolof en texte en utilisant l'alphabet latin.",
|
13 |
-
# input_label="Audio en wolof",
|
14 |
-
# output_label="Transcription alphabet latin"
|
15 |
-
# )
|
16 |
|
17 |
-
# demo.launch()
|
18 |
|
19 |
|
20 |
from speechbrain.inference.ASR import EncoderASR
|
21 |
import gradio as gr
|
22 |
import numpy as np
|
23 |
|
|
|
24 |
model = EncoderASR.from_hparams("speechbrain/asr-wav2vec2-dvoice-wolof")
|
25 |
|
|
|
26 |
def transcribe(audio):
|
27 |
sr, y = audio
|
28 |
y = y.astype(np.float32)
|
29 |
y /= np.max(np.abs(y))
|
30 |
|
31 |
-
|
|
|
32 |
|
33 |
-
|
34 |
-
|
35 |
-
|
36 |
-
|
37 |
-
"
|
|
|
38 |
)
|
39 |
|
40 |
-
|
|
|
41 |
|
|
|
1 |
+
# # from speechbrain.inference.ASR import EncoderASR
|
2 |
+
# # import gradio as gr
|
3 |
|
4 |
+
# # model = EncoderASR.from_hparams("speechbrain/asr-wav2vec2-dvoice-wolof")
|
5 |
|
6 |
+
# # def transcribe(audio):
|
7 |
+
# # return model.transcribe_file(audio.name)
|
8 |
|
9 |
|
10 |
+
# # demo = gr.Interface(fn=transcribe, inputs="file", outputs="text",
|
11 |
+
# # title="Transcription automatique du wolof",
|
12 |
+
# # description="Ce modèle transcrit un fichier audio en wolof en texte en utilisant l'alphabet latin.",
|
13 |
+
# # input_label="Audio en wolof",
|
14 |
+
# # output_label="Transcription alphabet latin"
|
15 |
+
# # )
|
16 |
|
17 |
+
# # demo.launch()
|
18 |
|
19 |
|
20 |
from speechbrain.inference.ASR import EncoderASR
|
21 |
import gradio as gr
|
22 |
import numpy as np
|
23 |
|
24 |
+
# Charger le modèle pré-entraîné
|
25 |
model = EncoderASR.from_hparams("speechbrain/asr-wav2vec2-dvoice-wolof")
|
26 |
|
27 |
+
# Définir la fonction de transcription
|
28 |
def transcribe(audio):
|
29 |
sr, y = audio
|
30 |
y = y.astype(np.float32)
|
31 |
y /= np.max(np.abs(y))
|
32 |
|
33 |
+
# Utiliser le modèle pour transcrire l'audio
|
34 |
+
return model.transcribe({"sampling_rate": sr, "raw": y})["text"]
|
35 |
|
36 |
+
# Créer l'interface Gradio avec le microphone et le téléchargement de fichier comme options d'entrée
|
37 |
+
asrdemo = gr.Interface(
|
38 |
+
title="Transcription audio en wolof latin by Papa Sega",
|
39 |
+
fn=transcribe,
|
40 |
+
inputs=gr.Audio(sources=["microphone", "upload"]),
|
41 |
+
outputs="text",
|
42 |
)
|
43 |
|
44 |
+
# Lancer l'application Gradio
|
45 |
+
asrdemo.launch()
|
46 |
|