Spaces:
Sleeping
Sleeping
Commit
•
a65ae92
1
Parent(s):
19047d4
Update app.py
Browse files
app.py
CHANGED
|
@@ -19,7 +19,11 @@ import numpy as np
|
|
| 19 |
|
| 20 |
from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
|
| 21 |
|
| 22 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
| 23 |
######################################################
|
| 24 |
def audio_to_text(audio, num_speakers):
|
| 25 |
path, error = convert_to_wav(audio)
|
|
@@ -31,12 +35,6 @@ def audio_to_text(audio, num_speakers):
|
|
| 31 |
return "Audio duration too long"
|
| 32 |
|
| 33 |
|
| 34 |
-
model = whisper.load_model("large-v2")
|
| 35 |
-
embedding_model = PretrainedSpeakerEmbedding(
|
| 36 |
-
"speechbrain/spkrec-ecapa-voxceleb",
|
| 37 |
-
device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
|
| 38 |
-
)
|
| 39 |
-
|
| 40 |
result = model.transcribe(path)
|
| 41 |
segments = result["segments"]
|
| 42 |
|
|
|
|
| 19 |
|
| 20 |
from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
|
| 21 |
|
| 22 |
+
model = whisper.load_model("large-v2")
|
| 23 |
+
embedding_model = PretrainedSpeakerEmbedding(
|
| 24 |
+
"speechbrain/spkrec-ecapa-voxceleb",
|
| 25 |
+
device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
|
| 26 |
+
)
|
| 27 |
######################################################
|
| 28 |
def audio_to_text(audio, num_speakers):
|
| 29 |
path, error = convert_to_wav(audio)
|
|
|
|
| 35 |
return "Audio duration too long"
|
| 36 |
|
| 37 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 38 |
result = model.transcribe(path)
|
| 39 |
segments = result["segments"]
|
| 40 |
|