vineelpratap
commited on
Commit
•
d4e3aaf
1
Parent(s):
b226b85
Update asr.py
Browse files
asr.py
CHANGED
@@ -66,10 +66,19 @@ model = Wav2Vec2ForCTC.from_pretrained(MODEL_ID)
|
|
66 |
# )
|
67 |
|
68 |
|
69 |
-
def transcribe(
|
70 |
-
|
71 |
-
):
|
72 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
73 |
|
74 |
lang_code = lang.split()[0]
|
75 |
processor.tokenizer.set_target_lang(lang_code)
|
|
|
66 |
# )
|
67 |
|
68 |
|
69 |
+
def transcribe(audio_data, lang="eng (English)"):
|
70 |
+
|
71 |
+
if isinstance(audio_data, tuple):
|
72 |
+
# microphone
|
73 |
+
sr, audio_samples = audio_data
|
74 |
+
print("case1", audio_samples[:5])
|
75 |
+
assert sr == ASR_SAMPLING_RATE, "Invalid sampling rate"
|
76 |
+
else:
|
77 |
+
# file upload
|
78 |
+
isinstance(audio_data, str)
|
79 |
+
print("case2 1", audio_data)
|
80 |
+
audio_samples = librosa.load(audio_fp, sr=ASR_SAMPLING_RATE, mono=True)[0]
|
81 |
+
print("case2", audio_samples[:5])
|
82 |
|
83 |
lang_code = lang.split()[0]
|
84 |
processor.tokenizer.set_target_lang(lang_code)
|