Spaces:
Running
Running
Update whisper_transcribe.py
Browse files- whisper_transcribe.py +1 -77
whisper_transcribe.py
CHANGED
|
@@ -1,6 +1,4 @@
|
|
| 1 |
import os
|
| 2 |
-
import sounddevice as sd
|
| 3 |
-
import soundfile as sf
|
| 4 |
import tempfile
|
| 5 |
import whisper
|
| 6 |
from utils import get_best_device
|
|
@@ -39,78 +37,4 @@ def transcribe_file(path: str, model_name: str = WHISPER_MODEL_NAME) -> str:
|
|
| 39 |
result = model.transcribe(path, fp16=False)
|
| 40 |
return result.get("text", "").strip()
|
| 41 |
except Exception as e:
|
| 42 |
-
raise RuntimeError(f"Whisper transcription failed: {e}")
|
| 43 |
-
|
| 44 |
-
|
| 45 |
-
def list_input_devices():
|
| 46 |
-
"""
|
| 47 |
-
Return a list of (index, name) for all input-capable audio devices.
|
| 48 |
-
"""
|
| 49 |
-
devices = sd.query_devices()
|
| 50 |
-
return [(i, d.get('name', '')) for i, d in enumerate(devices) if d.get('max_input_channels', 0) > 0]
|
| 51 |
-
|
| 52 |
-
|
| 53 |
-
def record_audio(duration: int, sr: int = 16000, device_index: int = None) -> tuple[int, list[float]]:
|
| 54 |
-
"""
|
| 55 |
-
Record `duration` seconds of mono audio at sample rate `sr`.
|
| 56 |
-
Automatically selects a valid input device if none is provided.
|
| 57 |
-
Raises RuntimeError on failure.
|
| 58 |
-
"""
|
| 59 |
-
devices = list_input_devices()
|
| 60 |
-
if not devices:
|
| 61 |
-
raise RuntimeError("No input audio devices found.")
|
| 62 |
-
|
| 63 |
-
# Choose device: user-specified or default, else first available
|
| 64 |
-
if device_index is None:
|
| 65 |
-
default = sd.default.device
|
| 66 |
-
if isinstance(default, (tuple, list)) and default[0] is not None:
|
| 67 |
-
device_index = default[0]
|
| 68 |
-
else:
|
| 69 |
-
device_index = devices[0][0]
|
| 70 |
-
|
| 71 |
-
# Validate device
|
| 72 |
-
if device_index not in [i for i, _ in devices]:
|
| 73 |
-
device_index = devices[0][0]
|
| 74 |
-
|
| 75 |
-
# Configure recorder
|
| 76 |
-
sd.default.device = (device_index, None)
|
| 77 |
-
sd.default.samplerate = sr
|
| 78 |
-
sd.default.channels = 1
|
| 79 |
-
|
| 80 |
-
print(f"Recording {duration}s at {sr}Hz from device #{device_index}…")
|
| 81 |
-
try:
|
| 82 |
-
data = sd.rec(int(duration * sr), dtype='float32', channels=1)
|
| 83 |
-
sd.wait()
|
| 84 |
-
except Exception as e:
|
| 85 |
-
raise RuntimeError(f"Audio recording failed: {e}")
|
| 86 |
-
|
| 87 |
-
# Convert to 1D float32 list
|
| 88 |
-
wav = data.flatten().tolist()
|
| 89 |
-
return sr, wav
|
| 90 |
-
|
| 91 |
-
|
| 92 |
-
def transcribe_from_mic(duration: int) -> str:
|
| 93 |
-
"""
|
| 94 |
-
Record `duration` seconds from the microphone, save to a temp WAV, transcribe, then clean up.
|
| 95 |
-
Raises RuntimeError on recording or transcription errors.
|
| 96 |
-
"""
|
| 97 |
-
try:
|
| 98 |
-
sr, wav = record_audio(duration)
|
| 99 |
-
except Exception as e:
|
| 100 |
-
raise RuntimeError(f"Microphone recording error: {e}")
|
| 101 |
-
|
| 102 |
-
# Write to temporary WAV file
|
| 103 |
-
with tempfile.NamedTemporaryFile(suffix=".wav", delete=False) as tmp:
|
| 104 |
-
tmp_path = tmp.name
|
| 105 |
-
sf.write(tmp_path, wav, sr)
|
| 106 |
-
|
| 107 |
-
try:
|
| 108 |
-
text = transcribe_file(tmp_path)
|
| 109 |
-
finally:
|
| 110 |
-
# Ensure cleanup
|
| 111 |
-
try:
|
| 112 |
-
os.remove(tmp_path)
|
| 113 |
-
except OSError:
|
| 114 |
-
pass
|
| 115 |
-
|
| 116 |
-
return text
|
|
|
|
| 1 |
import os
|
|
|
|
|
|
|
| 2 |
import tempfile
|
| 3 |
import whisper
|
| 4 |
from utils import get_best_device
|
|
|
|
| 37 |
result = model.transcribe(path, fp16=False)
|
| 38 |
return result.get("text", "").strip()
|
| 39 |
except Exception as e:
|
| 40 |
+
raise RuntimeError(f"Whisper transcription failed: {e}")
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|