WIP

app.py

@@ -45,8 +45,8 @@ def generate_srt(segments):
     """Generate SRT format subtitles from transcription segments."""
     srt_content = []
     for i, segment in enumerate(segments, 1):
-        start_time = format_time(segment["start"])
-        end_time = format_time(segment["end"])
+        start_time = format_time(segment["start_time"])
+        end_time = format_time(segment["end_time"])
         text = segment.get("text", "").strip()
         srt_content.append(f"{i}\n{start_time} --> {end_time}\n{text}\n\n")
     return "".join(srt_content)
@@ -223,7 +223,7 @@ def check_api_health():
         logger.error(f"Health check failed: {str(e)}")
         raise gr.Error("Failed to connect to the API service. Please try again later.")
 
-def transcribe_youtube(url, return_timestamps, generate_subs, chunk_length_s=30, batch_size=128):
+def transcribe_youtube(url, return_timestamps, generate_subs, chunk_length_s=15, batch_size=8):
     """Transcribe audio from YouTube video using URL endpoint."""
     logger.info(f"Starting YouTube transcription process for URL: {url}")
     logger.info(f"Parameters - return_timestamps: {return_timestamps}, generate_subs: {generate_subs}, chunk_length_s: {chunk_length_s}, batch_size: {batch_size}")
@@ -300,7 +300,7 @@ def transcribe_youtube(url, return_timestamps, generate_subs, chunk_length_s=30,
         logger.exception(f"Error in YouTube transcription: {str(e)}")
         raise gr.Error(f"Failed to transcribe YouTube video: {str(e)}")
 
-def transcribe(inputs, return_timestamps, generate_subs, chunk_length_s=30, batch_size=128):
+def transcribe(inputs, return_timestamps, generate_subs, chunk_length_s=15, batch_size=8):
     """Transcribe audio input using Whisper API."""
     logger.info(f"Starting transcription process for file: {inputs}")
     logger.info(f"Parameters - return_timestamps: {return_timestamps}, generate_subs: {generate_subs}, chunk_length_s: {chunk_length_s}, batch_size: {batch_size}")
@@ -385,7 +385,7 @@ youtube_transcribe = gr.Interface(
         gr.Checkbox(label="Include timestamps", value=True),
         gr.Checkbox(label="Generate subtitles", value=True),
         gr.Slider(minimum=5, maximum=30, value=15, step=5, label="Chunk Length (seconds)"),
-        gr.Slider(minimum=32, maximum=256, value=128, step=32, label="Batch Size")
+        gr.Slider(minimum=8, maximum=64, value=8, step=8, label="Batch Size")
     ],
     outputs=[
         gr.JSON(label="Transcription", open=True),
@@ -408,8 +408,8 @@ mf_transcribe = gr.Interface(
         gr.Audio(sources="microphone", type="filepath"),
         gr.Checkbox(label="Include timestamps", value=True),
         gr.Checkbox(label="Generate subtitles", value=True),
-        gr.Slider(minimum=5, maximum=30, value=30, step=5, label="Chunk Length (seconds)"),
-        gr.Slider(minimum=32, maximum=256, value=128, step=32, label="Batch Size")
+        gr.Slider(minimum=5, maximum=30, value=15, step=5, label="Chunk Length (seconds)"),
+        gr.Slider(minimum=8, maximum=64, value=8, step=8, label="Batch Size")
     ],
     outputs=[
         gr.JSON(label="Transcription", open=True),
@@ -430,7 +430,7 @@ file_transcribe = gr.Interface(
         gr.Checkbox(label="Include timestamps", value=True),
         gr.Checkbox(label="Generate subtitles", value=True),
         gr.Slider(minimum=5, maximum=30, value=15, step=5, label="Chunk Length (seconds)"),
-        gr.Slider(minimum=32, maximum=256, value=8, step=32, label="Batch Size")
+        gr.Slider(minimum=8, maximum=64, value=8, step=8, label="Batch Size")
     ],
     outputs=[
         gr.JSON(label="Transcription", open=True),