adding interface logic and audio returns

app.py

@@ -9,7 +9,7 @@ from lang_list import TEXT_SOURCE_LANGUAGE_NAMES , LANGUAGE_NAME_TO_CODE , text_
 from gradio_client import Client
 from dotenv import load_dotenv
 import requests
-from io import BytesIO
+from io import BytesIO  
 import cohere
 import os
 import re
@@ -56,9 +56,7 @@ def get_language_code(language_name):
 
 def translate_text(text, instructions=translatetextinst):
     """
-    - text (str): The initial text.
-    Returns:
-    - str: The translated text response.
+    translates text.
     """
     prompt = f"{text}{instructions}"
     response = co.generate(
@@ -117,12 +115,8 @@ class TaggedPhraseExtractor:
         for color, pattern in self.patterns.items():
             found_phrases = re.findall(pattern, self.text)
             sorted_phrases = sorted(found_phrases, key=len, reverse=True)
-            matches[color] = {
-                'all_phrases': found_phrases,
-                'top_three_longest': sorted_phrases[:3]
-            }
-            three_matches = sorted_phrases[:3]
-        return matches , three_matches
+            matches[color] = sorted_phrases[:3]
+        return matches
 
     def print_phrases(self):
         """Extract phrases and print them, including the three longest phrases."""
@@ -236,21 +230,26 @@ def process_input(image=None, file=None, audio=None, text="", translateto = "Eng
     processed_text = response.generations[0].text
 
     audio_output = process_text_to_audio(processed_text, translateto, translateto)
+    extractor = TaggedPhraseExtractor(final_text)
+    extractor.add_pattern("red", patterns["red"])
+    extractor.add_pattern("blue", patterns["blue"])
+    extractor.add_pattern("green", patterns["green"])
+    matches = extractor.extract_phrases()
 
-    extractor = TaggedPhraseExtractor(processed_text)
-    longest_phrases = extractor.get_longest_phrases()
+    top_phrases = []
+    for color, phrases in matches.items():
+        top_phrases.extend(phrases)
 
-    # Translate the longest phrases back into the native language
-    translated_phrases = [translate_text(phrase, translateto, translatefrom) for phrase in longest_phrases]
-
-    # Convert the original and translated phrases to audio
-    audio_samples = {
-        "target_language": [text_to_audio(phrase, translateto) for phrase in longest_phrases],
-        "native_language": [text_to_audio(phrase, translatefrom) for phrase in translated_phrases]
-    }
-
-    return audio_output, processed_text, audio_samples, longest_phrases, translated_phrases
+    audio_outputs = []
+    translations = []
+    for phrase in top_phrases:
+        translated_phrase = translate_text(phrase, translatefrom=translatefrom, translateto=translateto)
+        translations.append(translated_phrase)
+        target_audio = process_text_to_audio(phrase, translatefrom=translateto, translateto=translateto)
+        native_audio = process_text_to_audio(translated_phrase, translatefrom=translatefrom, translateto=translatefrom)
+        audio_outputs.append((target_audio, native_audio))
 
+    return final_text, audio_output, top_phrases, translations, audio_outputs
 
 def main():
     with gr.Blocks() as demo:
@@ -271,7 +270,7 @@ def main():
         
         process_button = gr.Button("🌟AyaTonic")
         
-        processed_text_output = RichTextbox(label="Processed Text")
+        processed_text_output = gr.RichTextbox(label="Processed Text")
         longest_phrases_1 = gr.Textbox(label="Focus")
         translated_phrases_output_1 = gr.Textbox(label="Translated Phrases")
         audio_output_native_phrase_1 = gr.Audio(label="Audio Output (Native Language)")
@@ -286,14 +285,42 @@ def main():
         audio_output_target_phrase_3 = gr.Audio(label="Audio Output (Target Language)")
         
         def update_outputs(image, file, audio, text, input_language, target_language):
-            processed_text, audio_samples, longest_phrases, translated_phrases = process_input(
-                image, file, audio, text, input_language, target_language)
-            return processed_text, audio_samples['native_language'], audio_samples['target_language'], "\n".join(longest_phrases), "\n".join(translated_phrases) # Fix this
-        
+            final_text, top_phrases, translations, audio_outputs = process_input(
+                image=image, file=file, audio=audio, text=text, 
+                translatefrom=input_language, translateto=target_language
+            )
+
+            # Prepare outputs for Gradio
+            processed_text_output = final_text
+            audio_output_native_phrases = [native for _, native in audio_outputs]
+            audio_output_target_phrases = [target for target, _ in audio_outputs]
+
+            # Assuming there are exactly 3 top phrases for simplicity
+            longest_phrases_outputs = top_phrases[:3]
+            translated_phrases_outputs = translations[:3]
+            audio_outputs_native = audio_output_native_phrases[:3]
+            audio_outputs_target = audio_output_target_phrases[:3]
+
+            return (
+                processed_text_output, 
+                *audio_outputs_native, 
+                *audio_outputs_target, 
+                *longest_phrases_outputs, 
+                *translated_phrases_outputs
+            )
+       
         process_button.click(
             fn=update_outputs,
             inputs=[image_input, file_input, audio_input, text_input, input_language, target_language],
-            outputs=[processed_text_output, audio_output_native_phrase_1, audio_output_target_phrase_1, longest_phrases_1, translated_phrases_output_1, audio_output_native_phrase_2, audio_output_target_phrase_2, longest_phrases_2, translated_phrases_output_2, audio_output_native_phrase_3, audio_output_target_phrase_3, longest_phrases_3, translated_phrases_output_3] #add education output
+            outputs=[
+                processed_text_output, 
+                audio_output_native_phrase_1, audio_output_target_phrase_1, 
+                audio_output_native_phrase_2, audio_output_target_phrase_2, 
+                audio_output_native_phrase_3, audio_output_target_phrase_3, 
+                longest_phrases_1, translated_phrases_output_1, 
+                longest_phrases_2, translated_phrases_output_2, 
+                longest_phrases_3, translated_phrases_output_3
+            ]
         )
 
 if __name__ == "__main__":

ayatonic.env

@@ -1,2 +0,0 @@
-CO_API_KEY=KQBPf0H0ENZESIC5nuUJ4i4jjg34xMPAkYK7s31W
-SEAMLESSM4T=https://facebook-seamless-m4t-v2-large.hf.space/--replicas/v4gsf/

requirements.txt

@@ -1,11 +1,11 @@
 gradio
 gradio_rich_textbox
 gradio-client
-cohere
-surya-ocr
-pillow
 torchvision
 torch
 python-dotenv
 pandas
-pydub
+pydub
+cohere
+surya-ocr
+pillow