Update app.py

app.py

@@ -3,10 +3,16 @@ from transformers import MarianMTModel, MarianTokenizer
 from datetime import datetime
 import langid
 import os
-import speech_recognition as sr
+import pyttsx3
+import time
 import warnings
 
+# Optional: suppress sacremoses warning
 warnings.filterwarnings("ignore", message="Recommended: pip install sacremoses.")
+
+# Set FFmpeg path explicitly (for pydub and audio playback)
+os.environ["PATH"] += os.pathsep + r"C:\ffmpeg\bin"
+
 langid.set_languages(['en', 'fr', 'sw'])
 
 MODEL_MAP = {
@@ -23,11 +29,6 @@ TONE_MODIFIERS = {
     "Casual": "Make this sound casual: "
 }
 
-VOICE_IDS = {
-    "Rachel (Female)": "21m00Tcm4TlvDq8ikWAM",
-    "Adam (Male)": "pNInz6obpgDQGcFmaJgB"
-}
-
 loaded_models = {}
 
 def load_model(model_name):
@@ -47,13 +48,15 @@ def detect_language(text):
 def translate(text, direction, tone):
     detected_lang = detect_language(text)
     expected_src = direction.split(" → ")[0].lower()
-    warning = ""
+
     if expected_src.startswith("english") and detected_lang != "en":
-        warning = f"⚠️ Detected language is '{detected_lang}', but you selected English as source."
+        warning = f"⚠ Detected language is '{detected_lang}', but you selected English as source."
     elif expected_src.startswith("french") and detected_lang != "fr":
-        warning = f"⚠️ Detected language is '{detected_lang}', but you selected French as source."
+        warning = f"⚠ Detected language is '{detected_lang}', but you selected French as source."
     elif expected_src.startswith("swahili") and detected_lang != "sw":
-        warning = f"⚠️ Detected language is '{detected_lang}', but you selected Swahili as source."
+        warning = f"⚠ Detected language is '{detected_lang}', but you selected Swahili as source."
+    else:
+        warning = ""
 
     prompt = TONE_MODIFIERS[tone] + text
     model_info = MODEL_MAP[direction]
@@ -80,22 +83,40 @@ def translate(text, direction, tone):
 
     return f"{warning}\n{translation}" if warning else translation
 
-# 🔇 TTS temporarily disabled
-def tts_via_api(text, voice_choice):
-    return None
+# Get available voices
+engine = pyttsx3.init()
+voices = engine.getProperty('voices')
+voice_names = [voice.name for voice in voices]
+
+def speak_text_to_file(text, voice_name):
+    try:
+        engine = pyttsx3.init()
+        engine.setProperty('rate', 150)
+        for voice in voices:
+            if voice.name == voice_name:
+                engine.setProperty('voice', voice.id)
+                break
+        output_path = "tts_output.wav"
+        engine.save_to_file(text, output_path)
+        engine.runAndWait()
+        return output_path
+    except Exception as e:
+        return None
 
 def transcribe_and_translate(audio_path, direction, tone):
+    import speech_recognition as sr
     recognizer = sr.Recognizer()
     try:
         with sr.AudioFile(audio_path) as source:
             audio = recognizer.record(source)
         if len(audio.frame_data) < 10000:
-            return "⚠️ Audio too short or empty. Please try again."
+            return "⚠ Audio too short or empty. Please try again."
         text = recognizer.recognize_google(audio)
         return translate(text, direction, tone)
     except Exception as e:
-        return f"⚠️ Could not transcribe audio: {e}"
+        return f"⚠ Could not transcribe audio: {e}"
 
+# Gradio UI
 with gr.Blocks(theme=gr.themes.Soft()) as demo:
     gr.Markdown("## 🌍 EAC Translator")
     gr.Markdown("Supports English, French, and Swahili. Includes tone control, language detection, voice input, and speech playback.")
@@ -106,33 +127,23 @@ with gr.Blocks(theme=gr.themes.Soft()) as demo:
             direction = gr.Dropdown(choices=list(MODEL_MAP.keys()), label="Translation Direction", value="English → Swahili")
             tone = gr.Radio(choices=list(TONE_MODIFIERS.keys()), label="Tone", value="Neutral")
             output_text = gr.Textbox(label="Translated Text", lines=3)
-            voice_select = gr.Dropdown(choices=list(VOICE_IDS.keys()), label="Voice", value="Rachel (Female)")
+            voice_choice = gr.Dropdown(choices=voice_names, label="Voice for Playback", value=voice_names[0])
+            audio_output = gr.Audio(label="Playback", interactive=False)
             translate_btn = gr.Button("Translate")
             speak_btn = gr.Button("🔊 Speak Translation")
-            audio_output = gr.Audio(label="Playback", interactive=False)
+            translate_btn.click(fn=translate, inputs=[input_text, direction, tone], outputs=output_text)
+            speak_btn.click(fn=speak_text_to_file, inputs=[output_text, voice_choice], outputs=audio_output)
 
-        with gr.Tab("🎙️ Voice Translation"):
+        with gr.Tab("🎙 Voice Translation"):
             audio_input = gr.Audio(sources=["microphone"], type="filepath", label="Speak Now")
             direction_voice = gr.Dropdown(choices=list(MODEL_MAP.keys()), label="Translation Direction", value="English → Swahili")
             tone_voice = gr.Radio(choices=list(TONE_MODIFIERS.keys()), label="Tone", value="Neutral")
             voice_output = gr.Textbox(label="Translated Text")
-            voice_select2 = gr.Dropdown(choices=list(VOICE_IDS.keys()), label="Voice", value="Rachel (Female)")
+            voice_choice2 = gr.Dropdown(choices=voice_names, label="Voice for Playback", value=voice_names[0])
+            audio_output2 = gr.Audio(label="Playback", interactive=False)
             voice_translate_btn = gr.Button("Transcribe & Translate")
             voice_speak_btn = gr.Button("🔊 Speak Translation")
-            audio_output2 = gr.Audio(label="Playback", interactive=False)
-
-    translate_btn.click(fn=translate, inputs=[input_text, direction, tone], outputs=output_text)
-    speak_btn.click(fn=tts_via_api, inputs=[output_text, voice_select], outputs=audio_output)
-    voice_translate_btn.click(fn=transcribe_and_translate, inputs=[audio_input, direction_voice, tone_voice], outputs=voice_output)
-    voice_speak_btn.click(fn=tts_via_api, inputs=[voice_output, voice_select2], outputs=audio_output2)
-
-    gr.Markdown(
-        """<div style='text-align: center;'>
-        <a href='https://eng-jobbers.vercel.app/' target='_blank' style='text-decoration: none; font-weight: bold;'>
-        Built with ❤️ by Eng. Jobbers – Qtrinova Inc
-        </a>
-        </div>""",
-        elem_id="footer"
-    )
+            voice_translate_btn.click(fn=transcribe_and_translate, inputs=[audio_input, direction_voice, tone_voice], outputs=voice_output)
+            voice_speak_btn.click(fn=speak_text_to_file, inputs=[voice_output, voice_choice2], outputs=audio_output2)
 
 demo.launch()