Update app.py

app.py

@@ -34,6 +34,12 @@ combined_models = []
 combined_models.extend(whisper_models)
 combined_models.extend(custom_models)
 
+usage = requests.get('https://api-free.deepl.com/v2/usage', headers=headers)
+usage = json.loads(usage.text)
+deepL_character_usage = str(usage['character_count'])
+print("deepL_character_usage")
+
+
 
 LANGUAGES = {
     "en": "English",
@@ -217,7 +223,7 @@ def speech_to_text(video_file_path, selected_source_lang, whisper_model):
     1. Download youtube video with a given url
     2. Watch it in the first video component
     3. Run automatic speech recognition on the video using fast Whisper models
-    4. Translate the recognized transcriptions to 26 languages supported by deepL
+    4. Translate the recognized transcriptions to 26 languages supported by deepL (If free API usage for the month is not yet fully consumed)
     5. Download generated subtitles in .vtt and .srt formats
     6. Watch the the original video with generated subtitles
     
@@ -229,13 +235,19 @@ def speech_to_text(video_file_path, selected_source_lang, whisper_model):
         raise ValueError("Error no video input")
     print(video_file_path)
     try:
+
+        
+
         _,file_ending = os.path.splitext(f'{video_file_path}')
         print(f'file enging is {file_ending}')
         print("starting conversion to wav")
         os.system(f'ffmpeg -i "{video_file_path}" -ar 16000 -ac 1 -c:a pcm_s16le "{video_file_path.replace(file_ending, ".wav")}"')
         print("conversion to wav ready")
+    
+    except Exception as e:
+        raise RuntimeError("Error Running inference with local model", e)
 
-
+    try:
 
         print("starting whisper c++")
         srt_path = str(video_file_path.replace(file_ending, ".wav")) + ".srt"
@@ -249,7 +261,7 @@ def speech_to_text(video_file_path, selected_source_lang, whisper_model):
                 os.system(f'./whisper.cpp/main "{video_file_path.replace(file_ending, ".wav")}" -t 4 -l {source_languages.get(selected_source_lang)} -m ./whisper.cpp/models/ggml-{whisper_model}.bin -osrt')
         print("starting whisper done with whisper")
     except Exception as e:
-        raise RuntimeError("Error converting video to audio")
+        raise RuntimeError("Error running Whisper cpp model")
 
     try:    
 
@@ -283,12 +295,27 @@ def speech_to_text(video_file_path, selected_source_lang, whisper_model):
             }
     
             df = pd.concat([df, pd.DataFrame(srt_to_df)])
+    except Exception as e:
+        print("Error creating srt df")
+
         
-                    
-        return df
+    try:
+        usage = requests.get('https://api-free.deepl.com/v2/usage', headers=headers)
+        usage = json.loads(usage.text)
+        char_count = str(usage['character_count'])
+    
+        print('Usage is at: ' + str(usage['character_count']) + ' characters')
+        
+        if usage['character_count'] >= 490000:
+            print("USAGE CLOSE TO LIMIT")
     
     except Exception as e:
-        raise RuntimeError("Error Running inference with local model", e)
+        print('Error with DeepL API requesting usage count')
+    
+                    
+    return df
+    
+
 
 
 def translate_transcriptions(df, selected_translation_lang_2):
@@ -316,20 +343,24 @@ def translate_transcriptions(df, selected_translation_lang_2):
         
         usage = requests.get('https://api-free.deepl.com/v2/usage', headers=headers)
         usage = json.loads(usage.text)
+        deepL_character_usage = str(usage['character_count'])
         try:
-            print('Usage is at: ' + str(usage['character_count']) + 'characters')
+            print('Usage is at: ' + deepL_character_usage + 'characters')
         except Exception as e:
             print(e)
         
-        if usage['character_count'] >= 490000:
-            print("USAGE CLOSE TO LIMIT")
-        
-        response = requests.post('https://api-free.deepl.com/v2/translate', headers=headers, data=data)
+        if int(deepL_character_usage) <= 490000:
+            print("STILL CHARACTERS LEFT")
+            response = requests.post('https://api-free.deepl.com/v2/translate', headers=headers, data=data)
     
-        # Print the response from the server
-        translated_sentences = json.loads(response.text)
-        translated_sentences = translated_sentences['translations'][0]['text'].split('\n')
-        df['translation'] = translated_sentences
+            # Print the response from the server
+            translated_sentences = json.loads(response.text)
+            translated_sentences = translated_sentences['translations'][0]['text'].split('\n')
+            df['translation'] = translated_sentences
+
+        else:
+            df['translation'] = df['text']    
+
     except Exception as e:
         print("EXCEPTION WITH DEEPL API")
         print(e)
@@ -391,7 +422,7 @@ def translate_transcriptions(df, selected_translation_lang_2):
         
     print("SRT DONE") 
     subtitle_files = ['subtitles.vtt','subtitles.srt']
-    
+
     return df, subtitle_files
 
 # def burn_srt_to_video(srt_file, video_in):
@@ -467,6 +498,10 @@ demo = gr.Blocks(css='''
 .output-markdown {max-width: 65ch !important;}
 ''')
 demo.encrypt = False
+
+
+
+
 with demo:
     transcription_var = gr.Variable()
     
@@ -484,8 +519,9 @@ with demo:
             
         with gr.Column():
             gr.Markdown('''
-            ### 1. Copy any Youtube video URL to box below 
-            (But please **consider using short videos** so others won't get queued) or click one of the examples and then press button "1. Download Youtube video"-button:
+             ### 1. Copy any non-private Youtube video URL to box below or click one of the examples.
+            (But please **consider using short videos** so others won't get queued) <br>
+            Then press button "1. Download Youtube video"-button:
             ''')
             examples = gr.Examples(examples=
                 [ "https://www.youtube.com/watch?v=nlMuHtV82q8&ab_channel=NothingforSale24", 
@@ -509,13 +545,13 @@ with demo:
             with gr.Column():
                 gr.Markdown('''
                 ##### Here you can start the transcription and translation process.
-                ##### Be aware that processing will last some time. With base model it is around 3x speed
-                ##### **Please select source language** for better transcriptions. Using 'Let the model analyze' makes mistakes sometimes and may lead to bad transcriptions
+                Be aware that processing will last some time. With base model it is around 3x speed
+                **Please select source language** for better transcriptions. Using 'Let the model analyze' makes mistakes sometimes and may lead to bad transcriptions
                 ''')
             selected_source_lang.render()
             selected_whisper_model.render()
             transcribe_btn = gr.Button("Step 2. Transcribe audio")
-            transcribe_btn.click(speech_to_text, [video_in, selected_source_lang, selected_whisper_model], transcription_df)
+            transcribe_btn.click(speech_to_text, [video_in, selected_source_lang, selected_whisper_model], [transcription_df])
 
             
     with gr.Row():
@@ -530,9 +566,15 @@ with demo:
     with gr.Row():
         with gr.Column():
             gr.Markdown('''
-            ##### PLEASE READ BELOW 
+            ### PLEASE READ BELOW
             Here you will can translate transcriptions to 26 languages.
-            If spoken language is not in the list, translation might not work. In this case original transcriptions are used
+            If spoken language is not in the list, translation might not work. In this case original transcriptions are used.
+            ''')
+            gr.Markdown(f'''
+            DeepL API character usage:
+            {deepL_character_usage if deepL_character_usage is not None else ''}/500 000 characters
+            If usage is over 490 000 characters original transcriptions will be used for subtitles.
+            API usage resets on 5th of every month.
             ''')
             selected_translation_lang_2.render()
             translate_transcriptions_button = gr.Button("Step 3. Translate transcription")