Spaces:
Sleeping
Sleeping
Madhuslista
commited on
Commit
·
8ee91c9
1
Parent(s):
138226e
Feature: Add time measurements
Browse files- lib/model.py +13 -0
lib/model.py
CHANGED
@@ -2,6 +2,7 @@
|
|
2 |
# -*- coding: utf-8 -*-
|
3 |
|
4 |
import gc
|
|
|
5 |
import torch
|
6 |
import whisperx as wx
|
7 |
|
@@ -27,13 +28,25 @@ def transcribe_audio(audio_file, audio_path, transcript_folder_path):
|
|
27 |
# Transcribe the audio
|
28 |
print("Starting transcription...")
|
29 |
print("Loading model...")
|
|
|
30 |
model = wx.load_model("large-v2", device=DEVICE, compute_type=COMPUTE_TYPE, language="en")
|
|
|
31 |
print("Loading audio...")
|
|
|
32 |
audio = wx.load_audio(audio_path)
|
|
|
33 |
print("Transcribing...")
|
|
|
34 |
result = model.transcribe(audio, batch_size=BATCH_SIZE)
|
|
|
35 |
print("Transcription complete!")
|
36 |
|
|
|
|
|
|
|
|
|
|
|
|
|
37 |
# Save the transcript to a file
|
38 |
text = "\n ".join([i["text"] for i in result["segments"]])
|
39 |
|
|
|
2 |
# -*- coding: utf-8 -*-
|
3 |
|
4 |
import gc
|
5 |
+
from time import time
|
6 |
import torch
|
7 |
import whisperx as wx
|
8 |
|
|
|
28 |
# Transcribe the audio
|
29 |
print("Starting transcription...")
|
30 |
print("Loading model...")
|
31 |
+
time_1 = time()
|
32 |
model = wx.load_model("large-v2", device=DEVICE, compute_type=COMPUTE_TYPE, language="en")
|
33 |
+
time_2 = time()
|
34 |
print("Loading audio...")
|
35 |
+
time_3 = time()
|
36 |
audio = wx.load_audio(audio_path)
|
37 |
+
time_4 = time()
|
38 |
print("Transcribing...")
|
39 |
+
time_5 = time()
|
40 |
result = model.transcribe(audio, batch_size=BATCH_SIZE)
|
41 |
+
time_6 = time()
|
42 |
print("Transcription complete!")
|
43 |
|
44 |
+
print("\nTime Report: ")
|
45 |
+
print("Loading model: ", round(time_2 - time_1,2), " [s]")
|
46 |
+
print("Loading audio: ", round(time_4 - time_3,2), " [s]")
|
47 |
+
print("Transcribing: ", round(time_6 - time_5,2), " [s]")
|
48 |
+
print("Total: ", round(time_6 - time_1,2), " [s]")
|
49 |
+
|
50 |
# Save the transcript to a file
|
51 |
text = "\n ".join([i["text"] for i in result["segments"]])
|
52 |
|