Merge branch 'main' of https://huggingface.co/spaces/aadnk/whisper-webui into main
Browse files- src/vad.py +6 -1
- src/whisper/fasterWhisperContainer.py +3 -1
src/vad.py
CHANGED
@@ -205,11 +205,16 @@ class AbstractTranscription(ABC):
|
|
205 |
|
206 |
print("Running whisper from ", format_timestamp(segment_start), " to ", format_timestamp(segment_end), ", duration: ",
|
207 |
segment_duration, "expanded: ", segment_expand_amount, "prompt: ", segment_prompt, "language: ", detected_language)
|
208 |
-
|
|
|
|
|
209 |
scaled_progress_listener = SubTaskProgressListener(progressListener, base_task_total=progress_total_duration,
|
210 |
sub_task_start=segment_start - progress_start_offset, sub_task_total=segment_duration)
|
211 |
segment_result = whisperCallable.invoke(segment_audio, segment_index, segment_prompt, detected_language, progress_listener=scaled_progress_listener)
|
212 |
|
|
|
|
|
|
|
213 |
adjusted_segments = self.adjust_timestamp(segment_result["segments"], adjust_seconds=segment_start, max_source_time=segment_duration)
|
214 |
|
215 |
# Propagate expand amount to the segments
|
|
|
205 |
|
206 |
print("Running whisper from ", format_timestamp(segment_start), " to ", format_timestamp(segment_end), ", duration: ",
|
207 |
segment_duration, "expanded: ", segment_expand_amount, "prompt: ", segment_prompt, "language: ", detected_language)
|
208 |
+
|
209 |
+
perf_start_time = time.perf_counter()
|
210 |
+
|
211 |
scaled_progress_listener = SubTaskProgressListener(progressListener, base_task_total=progress_total_duration,
|
212 |
sub_task_start=segment_start - progress_start_offset, sub_task_total=segment_duration)
|
213 |
segment_result = whisperCallable.invoke(segment_audio, segment_index, segment_prompt, detected_language, progress_listener=scaled_progress_listener)
|
214 |
|
215 |
+
perf_end_time = time.perf_counter()
|
216 |
+
print("Whisper took {} seconds".format(perf_end_time - perf_start_time))
|
217 |
+
|
218 |
adjusted_segments = self.adjust_timestamp(segment_result["segments"], adjust_seconds=segment_start, max_source_time=segment_duration)
|
219 |
|
220 |
# Propagate expand amount to the segments
|
src/whisper/fasterWhisperContainer.py
CHANGED
@@ -7,6 +7,7 @@ from src.hooks.progressListener import ProgressListener
|
|
7 |
from src.languages import get_language_from_name
|
8 |
from src.modelCache import ModelCache
|
9 |
from src.whisper.abstractWhisperContainer import AbstractWhisperCallback, AbstractWhisperContainer
|
|
|
10 |
|
11 |
class FasterWhisperContainer(AbstractWhisperContainer):
|
12 |
def __init__(self, model_name: str, device: str = None, compute_type: str = "float16",
|
@@ -138,7 +139,8 @@ class FasterWhisperCallback(AbstractWhisperCallback):
|
|
138 |
if progress_listener is not None:
|
139 |
progress_listener.on_progress(segment.end, info.duration)
|
140 |
if verbose:
|
141 |
-
print(segment.
|
|
|
142 |
|
143 |
text = " ".join([segment.text for segment in segments])
|
144 |
|
|
|
7 |
from src.languages import get_language_from_name
|
8 |
from src.modelCache import ModelCache
|
9 |
from src.whisper.abstractWhisperContainer import AbstractWhisperCallback, AbstractWhisperContainer
|
10 |
+
from src.utils import format_timestamp
|
11 |
|
12 |
class FasterWhisperContainer(AbstractWhisperContainer):
|
13 |
def __init__(self, model_name: str, device: str = None, compute_type: str = "float16",
|
|
|
139 |
if progress_listener is not None:
|
140 |
progress_listener.on_progress(segment.end, info.duration)
|
141 |
if verbose:
|
142 |
+
print("[{}->{}] {}".format(format_timestamp(segment.start, True), format_timestamp(segment.end, True),
|
143 |
+
segment.text))
|
144 |
|
145 |
text = " ".join([segment.text for segment in segments])
|
146 |
|