up

README.md

@@ -4,7 +4,7 @@ emoji: 🤫
 colorFrom: indigo
 colorTo: red
 sdk: gradio
-sdk_version: 3.9.1
+sdk_version: 4.16.0
 app_file: app.py
 pinned: false
 tags:

run_demo_ct2.py

@@ -42,6 +42,8 @@ GEN_KWARGS = {
     # "logprob_threshold": None,
     # vad threshold
     # "no_speech_threshold": None,
+    # "condition_on_previous_text": False,  # todo: only for distilled version
+    "vad_filter": True,
 }
 
 logging.basicConfig(
@@ -156,23 +158,24 @@ def infer(model, filename, with_timestamps, return_df=False):
             return text
 
 
-def transcribe(microphone, file_upload, with_timestamps, model_name=DEFAULT_MODEL_NAME):
-    warn_output = ""
-    if (microphone is not None) and (file_upload is not None):
-        warn_output = (
-            "WARNING: You've uploaded an audio file and used the microphone. "
-            "The recorded file from the microphone will be used and the uploaded audio will be discarded.\n"
-        )
+# def transcribe(microphone, file_upload, with_timestamps, model_name=DEFAULT_MODEL_NAME):
+def transcribe(audio_file_path, with_timestamps, model_name=DEFAULT_MODEL_NAME):
+    # warn_output = ""
+    # if (microphone is not None) and (file_upload is not None):
+    #     warn_output = (
+    #         "WARNING: You've uploaded an audio file and used the microphone. "
+    #         "The recorded file from the microphone will be used and the uploaded audio will be discarded.\n"
+    #     )
 
-    elif (microphone is None) and (file_upload is None):
-        return "ERROR: You have to either use the microphone or upload an audio file"
+    # elif (microphone is None) and (file_upload is None):
+    #     return "ERROR: You have to either use the microphone or upload an audio file"
 
-    file = microphone if microphone is not None else file_upload
+    # audio_file_path = microphone if microphone is not None else file_upload
 
     model = maybe_load_cached_pipeline(model_name)
-    # text = model.transcribe(file, **GEN_KWARGS)["text"]
-    # text = infer(model, file, with_timestamps)
-    text = infer(model, file, with_timestamps, return_df=True)
+    # text = model.transcribe(audio_file_path, **GEN_KWARGS)["text"]
+    # text = infer(model, audio_file_path, with_timestamps)
+    text = infer(model, audio_file_path, with_timestamps, return_df=True)
 
     logger.info(f'Transcription by `{model_name}`:\n{text.to_json(orient="index", force_ascii=False, indent=2)}\n')
 
@@ -233,8 +236,9 @@ with gr.Blocks() as demo:
           """
         )
 
-        microphone_input = gr.inputs.Audio(source="microphone", type="filepath", label="Record", optional=True)
-        upload_input = gr.inputs.Audio(source="upload", type="filepath", label="Upload File", optional=True)
+        # microphone_input = gr.Audio(sources="microphone", type="filepath", label="Record")
+        # upload_input = gr.Audio(sources="upload", type="filepath", label="Upload File")
+        audio_file_path = gr.Audio(sources=["microphone", "upload"], type="filepath", label="Record or upload file")
         with_timestamps_input = gr.Checkbox(label="With timestamps?")
 
         microphone_transcribe_btn = gr.Button("Transcribe Audio")
@@ -247,14 +251,11 @@ with gr.Blocks() as demo:
         text_output_df2 = gr.DataFrame(
             value=default_text_output_df,
             label="Transcription",
-            row_count=(0, "dynamic"),
-            max_rows=10,
             wrap=True,
-            overflow_row_behaviour="paginate",
         )
 
         microphone_transcribe_btn.click(
-            transcribe, inputs=[microphone_input, upload_input, with_timestamps_input], outputs=text_output_df2
+            transcribe, inputs=[audio_file_path, with_timestamps_input], outputs=text_output_df2
         )
 
     # with gr.Tab("Transcribe YouTube"):
@@ -301,7 +302,7 @@ with gr.Blocks() as demo:
           """
         )
 
-        yt_link_input = gr.inputs.Textbox(lines=1, placeholder="Paste the URL to a YouTube video here", label="YouTube URL")
+        yt_link_input = gr.Textbox(lines=1, placeholder="Paste the URL to a YouTube video here", label="YouTube URL")
         download_youtube_btn = gr.Button("Download Youtube video")
         downloaded_video_output = gr.Video(label="Video file", mirror_webcam=False)
         download_youtube_btn.click(download_video_from_youtube, inputs=[yt_link_input], outputs=[downloaded_video_output])
@@ -311,14 +312,10 @@ with gr.Blocks() as demo:
         text_output_df = gr.DataFrame(
             value=default_text_output_df,
             label="Transcription",
-            row_count=(0, "dynamic"),
-            max_rows=10,
             wrap=True,
-            overflow_row_behaviour="paginate",
         )
 
         video_transcribe_btn.click(video_transcribe, inputs=[downloaded_video_output, with_timestamps_input3], outputs=[text_output_df])
 
-# demo.launch(server_name="0.0.0.0", debug=True)
-# demo.launch(server_name="0.0.0.0", debug=True, share=True)
-demo.launch(enable_queue=True)
+# demo.queue(max_size=10).launch(server_name="0.0.0.0", debug=True, ssl_certfile="/home/bhuang/tools/cert.pem", ssl_keyfile="/home/bhuang/tools/key.pem", ssl_verify=False)
+demo.queue(max_size=10).launch()