Update app.py

app.py

@@ -21,6 +21,9 @@ import pandas as pd
 import re
 import time
 
+import subprocess
+import shlex
+
 from pytube import YouTube
 import torch
 
@@ -28,11 +31,13 @@ INTRO_MSG = '''
 #### <p>There are many not very widely spoken languages for which it is quite hard to find learning materials, 
 especially well dubbed videos (target language video with target language subs). 
 This tool will hopefully transcribe and add subs to your videos. 
-At least for me this is a nice tool to practice both listening and reading skills. 
+At least for me this is a nice tool to practice both listening and reading skills.
+This is a 'one-click' variant of similar spaces found here on the HF hub.
 <p>Speech Recognition is based on models from OpenAI Whisper - https://github.com/openai/whisper
 <p> This space is using the c++ implementation by https://github.com/ggerganov/whisper.cpp 
 '''
 
+
 whisper_models = MODELS_TO_DOWNLOAD #["medium"]#["base", "small", "medium", "large", "base.en"]
 
 custom_models = []
@@ -42,6 +47,104 @@ combined_models.extend(custom_models)
 
 LANGUAGES = {
     "bg": "Bulgarian",
+    "en": "English",
+    "zh": "Chinese",
+    "de": "German",
+    "es": "Spanish",
+    "ru": "Russian",
+    "ko": "Korean",
+    "fr": "French",
+    "ja": "Japanese",
+    "pt": "Portuguese",
+    "tr": "Turkish",
+    "pl": "Polish",
+    "ca": "Catalan",
+    "nl": "Dutch",
+    "ar": "Arabic",
+    "sv": "Swedish",
+    "it": "Italian",
+    "id": "Indonesian",
+    "hi": "Hindi",
+    "fi": "Finnish",
+    "vi": "Vietnamese",
+    "he": "Hebrew",
+    "uk": "Ukrainian",
+    "el": "Greek",
+    "ms": "Malay",
+    "cs": "Czech",
+    "ro": "Romanian",
+    "da": "Danish",
+    "hu": "Hungarian",
+    "ta": "Tamil",
+    "no": "Norwegian",
+    "th": "Thai",
+    "ur": "Urdu",
+    "hr": "Croatian",
+    "lt": "Lithuanian",
+    "la": "Latin",
+    "mi": "Maori",
+    "ml": "Malayalam",
+    "cy": "Welsh",
+    "sk": "Slovak",
+    "te": "Telugu",
+    "fa": "Persian",
+    "lv": "Latvian",
+    "bn": "Bengali",
+    "sr": "Serbian",
+    "az": "Azerbaijani",
+    "sl": "Slovenian",
+    "kn": "Kannada",
+    "et": "Estonian",
+    "mk": "Macedonian",
+    "br": "Breton",
+    "eu": "Basque",
+    "is": "Icelandic",
+    "hy": "Armenian",
+    "ne": "Nepali",
+    "mn": "Mongolian",
+    "bs": "Bosnian",
+    "kk": "Kazakh",
+    "sq": "Albanian",
+    "sw": "Swahili",
+    "gl": "Galician",
+    "mr": "Marathi",
+    "pa": "Punjabi",
+    "si": "Sinhala",
+    "km": "Khmer",
+    "sn": "Shona",
+    "yo": "Yoruba",
+    "so": "Somali",
+    "af": "Afrikaans",
+    "oc": "Occitan",
+    "ka": "Georgian",
+    "be": "Belarusian",
+    "tg": "Tajik",
+    "sd": "Sindhi",
+    "gu": "Gujarati",
+    "am": "Amharic",
+    "yi": "Yiddish",
+    "lo": "Lao",
+    "uz": "Uzbek",
+    "fo": "Faroese",
+    "ht": "Haitian creole",
+    "ps": "Pashto",
+    "tk": "Turkmen",
+    "nn": "Nynorsk",
+    "mt": "Maltese",
+    "sa": "Sanskrit",
+    "lb": "Luxembourgish",
+    "my": "Myanmar",
+    "bo": "Tibetan",
+    "tl": "Tagalog",
+    "mg": "Malagasy",
+    "as": "Assamese",
+    "tt": "Tatar",
+    "haw": "Hawaiian",
+    "ln": "Lingala",
+    "ha": "Hausa",
+    "ba": "Bashkir",
+    "jw": "Javanese",
+    "su": "Sundanese",
 }
 
 # language code lookup by name, with a few language aliases
@@ -60,7 +163,27 @@ def get_youtube(video_url):
     print(f"Download complete - {abs_video_path}")
     return abs_video_path
 
-def speech_to_text(video_file_path, selected_source_lang, whisper_model):
+def run_command(command, app_state):
+    print(command)
+    process = subprocess.Popen(command, shell=True, stdout=subprocess.PIPE)
+    while process.poll() is None:
+        time.sleep(5)
+        output = process.stdout.readline()
+        if output == '' and process.poll() is not None:
+            break
+        if output:
+            decoded = output.decode()
+            print(decoded)
+            app_state['output'] += decoded
+
+    rc = process.poll()
+    print(f'{cmd} ret code is {rc}')
+    return rc
+
+def speech_to_text(video_file_path, 
+                   selected_source_lang, 
+                   whisper_model,
+                   app_state):
     """    
     Speech Recognition is based on models from OpenAI Whisper https://github.com/openai/whisper
     This space is using c++ implementation by https://github.com/ggerganov/whisper.cpp
@@ -90,8 +213,9 @@ def speech_to_text(video_file_path, selected_source_lang, whisper_model):
     try:
         print("starting whisper c++")
         os.system(f'rm -f {srt_path}')
-        print('Running regular model')
-        os.system(f'./whisper.cpp/main "{input_wav_file}" -t {os.cpu_count()} -l {source_languages.get(selected_source_lang)} -m ./whisper.cpp/models/ggml-{whisper_model}.bin -osrt -ovtt')
+        run_command(f'./whisper.cpp/main "{input_wav_file}" -t {os.cpu_count()} -l {source_languages.get(selected_source_lang)} -m ./whisper.cpp/models/ggml-{whisper_model}.bin -osrt -ovtt',
+                   app_state)
+#         os.system(f'./whisper.cpp/main "{input_wav_file}" -t {os.cpu_count()} -l {source_languages.get(selected_source_lang)} -m ./whisper.cpp/models/ggml-{whisper_model}.bin -osrt -ovtt')
         print("whisper c++ finished")
     except Exception as e:
         raise RuntimeError("Error running Whisper cpp model")
@@ -149,6 +273,7 @@ subtitle_files = gr.File(
 video_player = gr.HTML('<p>video will be played here')
 eventslider = gr.Slider(visible=False)
 status_msg = gr.Markdown('Status')
+output_label = gr.Textbox('', interactive=False, show_label=False)
 
 demo = gr.Blocks()
 demo.encrypt = False
@@ -157,24 +282,26 @@ def set_app_msg(app_state, msg):
     app_state['status_msg'] = msg
 
 def transcribe(app_state, youtube_url_in, selected_source_lang, selected_whisper_model):
+    app_state['output'] = ''
     set_app_msg(app_state, 'Downloading the movie ...')
     video_file_path = get_youtube(youtube_url_in)
     set_app_msg(app_state, f'Running the speech to text model {selected_source_lang}/{selected_whisper_model}. This can take some time.')
-    subtitle_files = speech_to_text(video_file_path, selected_source_lang, selected_whisper_model)
+    subtitle_files = speech_to_text(video_file_path, selected_source_lang, selected_whisper_model, app_state)
     set_app_msg(app_state, f'Creating the video player ...')
     video_player = create_video_player(subtitle_files, video_file_path)
-    set_app_msg(app_state, f'Transcribing done, generating video player ...')
+    set_app_msg(app_state, f'Transcribing done, generating video player')
     return subtitle_files, video_player
     
 
 def on_change_event(app_state):
-    print('Running!')
-    return app_state['status_msg']
+    print(f'Running! {app_state}')
+    return app_state['status_msg'], app_state['output']
 
 with demo:
     app_state = gr.State({
-                          'running':False,
-                          'status_msg': ''
+                          'running': False,
+                          'status_msg': '',
+                          'output': ''
                          })
     
     with gr.Row():
@@ -196,12 +323,13 @@ with demo:
             
             eventslider.render()
             status_msg.render()
+            output_label.render()
             subtitle_files.render()
             video_player.render()
     with gr.Row():
         gr.Markdown('This app is based on [this code](https://huggingface.co/spaces/RASMUS/Whisper-youtube-crosslingual-subtitles/tree/main) by RASMUS.')
 
-    dep = demo.load(on_change_event, inputs=[app_state], outputs=[status_msg], every=10)
+    dep = demo.load(on_change_event, inputs=[app_state], outputs=[status_msg, output_label], every=10)
 
 
 #### RUN ###