Spaces:
Running
on
Zero
Running
on
Zero
Update app.py
Browse files
app.py
CHANGED
@@ -40,6 +40,7 @@ revision = None
|
|
40 |
#model_checkpoint = "oza75/bm-whisper-large-turbo-v4"
|
41 |
#model_checkpoint = "oza75/bm-whisper-large-v3-base"
|
42 |
model_checkpoint = "oza75/bm-whisper-large-v3-sft-3"
|
|
|
43 |
# language = "sundanese"
|
44 |
|
45 |
# Load the custom tokenizer designed for Bambara and the ASR model
|
@@ -47,7 +48,7 @@ model_checkpoint = "oza75/bm-whisper-large-v3-sft-3"
|
|
47 |
tokenizer = WhisperTokenizer.from_pretrained(model_checkpoint, device=device)
|
48 |
pipe = pipeline("automatic-speech-recognition", model=model_checkpoint, tokenizer=tokenizer, device=device, revision=revision)
|
49 |
|
50 |
-
LANGUAGES = {"bambara": "
|
51 |
|
52 |
def resample_audio(audio_path, target_sample_rate=16000):
|
53 |
"""
|
|
|
40 |
#model_checkpoint = "oza75/bm-whisper-large-turbo-v4"
|
41 |
#model_checkpoint = "oza75/bm-whisper-large-v3-base"
|
42 |
model_checkpoint = "oza75/bm-whisper-large-v3-sft-3"
|
43 |
+
model_checkpoint = "djelia/djelia-whisper-v2"
|
44 |
# language = "sundanese"
|
45 |
|
46 |
# Load the custom tokenizer designed for Bambara and the ASR model
|
|
|
48 |
tokenizer = WhisperTokenizer.from_pretrained(model_checkpoint, device=device)
|
49 |
pipe = pipeline("automatic-speech-recognition", model=model_checkpoint, tokenizer=tokenizer, device=device, revision=revision)
|
50 |
|
51 |
+
LANGUAGES = {"bambara": "french", "french": "french", "english": "english"}
|
52 |
|
53 |
def resample_audio(audio_path, target_sample_rate=16000):
|
54 |
"""
|