Update asr.py
Browse files
asr.py
CHANGED
@@ -3,16 +3,22 @@ from transformers import Wav2Vec2ForCTC, AutoProcessor
|
|
3 |
import torch
|
4 |
|
5 |
ASR_SAMPLING_RATE = 16_000
|
6 |
-
MODEL_ID = "facebook/mms-1b-all"
|
7 |
|
|
|
8 |
processor = AutoProcessor.from_pretrained(MODEL_ID)
|
9 |
model = Wav2Vec2ForCTC.from_pretrained(MODEL_ID)
|
10 |
|
11 |
-
def transcribe(
|
12 |
-
if
|
|
|
13 |
return "ERROR: You have to either use the microphone or upload an audio file"
|
14 |
|
15 |
-
audio_samples = librosa.load(
|
|
|
|
|
|
|
|
|
|
|
16 |
inputs = processor(audio_samples, sampling_rate=ASR_SAMPLING_RATE, return_tensors="pt")
|
17 |
|
18 |
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
|
@@ -21,12 +27,7 @@ def transcribe(audio):
|
|
21 |
|
22 |
with torch.no_grad():
|
23 |
outputs = model(**inputs).logits
|
|
|
|
|
24 |
|
25 |
-
# Setting the target language to Faroese (ISO 639-3: fao)
|
26 |
-
processor.tokenizer.set_target_lang("fao")
|
27 |
-
model.load_adapter("fao")
|
28 |
-
|
29 |
-
ids = torch.argmax(outputs, dim=-1)[0]
|
30 |
-
transcription = processor.decode(ids)
|
31 |
-
|
32 |
return transcription
|
|
|
3 |
import torch
|
4 |
|
5 |
ASR_SAMPLING_RATE = 16_000
|
|
|
6 |
|
7 |
+
MODEL_ID = "facebook/mms-1b-all"
|
8 |
processor = AutoProcessor.from_pretrained(MODEL_ID)
|
9 |
model = Wav2Vec2ForCTC.from_pretrained(MODEL_ID)
|
10 |
|
11 |
+
def transcribe(audio_source=None, microphone=None, file_upload=None):
|
12 |
+
audio_fp = file_upload if file_upload else microphone
|
13 |
+
if audio_fp is None:
|
14 |
return "ERROR: You have to either use the microphone or upload an audio file"
|
15 |
|
16 |
+
audio_samples = librosa.load(audio_fp, sr=ASR_SAMPLING_RATE, mono=True)[0]
|
17 |
+
|
18 |
+
# Set Faroese language
|
19 |
+
processor.tokenizer.set_target_lang("fao")
|
20 |
+
model.load_adapter("fao")
|
21 |
+
|
22 |
inputs = processor(audio_samples, sampling_rate=ASR_SAMPLING_RATE, return_tensors="pt")
|
23 |
|
24 |
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
|
|
|
27 |
|
28 |
with torch.no_grad():
|
29 |
outputs = model(**inputs).logits
|
30 |
+
ids = torch.argmax(outputs, dim=-1)[0]
|
31 |
+
transcription = processor.decode(ids)
|
32 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
33 |
return transcription
|