sbv -> srt

app.py

@@ -5,7 +5,7 @@ from pyannote.audio import Pipeline
 from transformers import pipeline
 from utils import second_to_timecode, download_from_youtube
 
-MODEL_NAME = 'bayartsogt/whisper-medium-mn-10'
+MODEL_NAME = 'bayartsogt/whisper-large-v2-mn-13'
 lang = 'mn'
 
 chunk_length_s = 9
@@ -60,21 +60,24 @@ def generator(youtube_link, microphone, file_upload, num_speakers, max_duration,
                 vad_start = _start + int(sampling_rate * vad_turn.start)
                 vad_end = _start + int(sampling_rate * vad_turn.end)
                 prediction = pipe(waveform[vad_start: vad_end])['text']
-                history +=  f"{second_to_timecode(speech_turn.start + vad_turn.start)},{second_to_timecode(speech_turn.start + vad_turn.end)}\n" + \
+                history +=  f"{counter}\n" + \
+                            f"{second_to_timecode(speech_turn.start + vad_turn.start)} --> {second_to_timecode(speech_turn.start + vad_turn.end)}\n" + \
                             f"{prediction}\n\n"
                             # f">> {speaker}: {prediction}\n\n"
                 yield history, history, None
+                counter += 1
 
         else:
             prediction = pipe(data)['text']
-            history +=  f"{second_to_timecode(speech_turn.start)},{second_to_timecode(speech_turn.end)}\n" + \
+            history +=  f"{counter}\n" + \
+                        f"{second_to_timecode(speech_turn.start)} --> {second_to_timecode(speech_turn.end)}\n" + \
                         f"{prediction}\n\n"
                         # f">> {speaker}: {prediction}\n\n"
-
-        yield history, history, None
+            counter += 1
+            yield history, history, None
     
     # https://support.google.com/youtube/answer/2734698?hl=en#zippy=%2Cbasic-file-formats%2Csubrip-srt-example%2Csubviewer-sbv-example
-    file_name = 'transcript.sbv'
+    file_name = 'transcript.srt'
     with open(file_name, 'w') as fp:
         fp.write(history)
     
@@ -96,7 +99,7 @@ demo = gr.Interface(
     title="Transcribe Mongolian Whisper 🇲🇳",
     description=(
         "Transcribe Youtube Video / Microphone / Uploaded File in Mongolian Whisper Model." + \
-        " | You can upload SubView file (`.sbv`) [to your youtube video](https://support.google.com/youtube/answer/2734698?hl=en#zippy=%2Cbasic-file-formats)." + \
+        " | You can upload SubRip file (`.srt`) [to your youtube video](https://support.google.com/youtube/answer/2734698?hl=en#zippy=%2Cbasic-file-formats)." + \
         " | Please REFRESH 🔄 the page after you transcribed!" + \
         " | 🐦 [@_tsogoo_](https://twitter.com/_tsogoo_)" + \
         " | 🤗 [@bayartsogt](https://huggingface.co/bayartsogt)" + \