Update app.py

app.py

@@ -18,10 +18,13 @@ def install_fairseq():
         subprocess.check_call(["pip", "install", "fairseq"])
         subprocess.check_call(["pip", "install", "sentencepiece"])
         subprocess.check_call(["pip", "install", "soundfile"])
+        subprocess.check_call(["pip", "install", "gTTS"])
         return "fairseq successfully installed!"
     except subprocess.CalledProcessError as e:
         return f"An error occurred while installing fairseq: {str(e)}"
 
+from gtts import gTTS
+
 def convert_audio_to_16k_wav(audio_input):
     sound = AudioSegment.from_file(audio_input)
     sample_rate = sound.frame_rate
@@ -48,30 +51,37 @@ def run_my_code(input_text, language):
     data_root=""
     model_checkpoint=""
     d_r=""
+    lang=''
 
     if(language=="Hindi"):
         model_checkpoint = "./models/hi_m.pt"
         data_root="./lang/hi/"
+        lang='hi'
 
     if(language=="Gujrati"):
         model_checkpoint = "./models/gj_m.pt"
         data_root="./lang/gj/"
+        lang='gu'
 
     if(language=="Bengali"):
         model_checkpoint = "./models/bn_m.pt"
-        data_root="./lang/bn/"        
+        data_root="./lang/bn/"      
+        lang='bn'  
 
     if(language=="Nepali"):
         model_checkpoint = "./models/ne_m.pt"
         data_root="./lang/ne/"
+        lang='ne'
 
     if(language=="Tamil"):
         model_checkpoint = "./models/tm_m.pt"
         data_root="./lang/tm/"
+        lang='ta'
 
     if(language=="Marathi"):
         model_checkpoint = "./models/mt_m.pt"
-        data_root="./lang/mt/"        
+        data_root="./lang/mt/" 
+        lang='mr'       
         
     #os.system(f"cp {hi_wav} {data_root}data/tst-COMMON/wav/test.wav")
     f = open('input.txt', 'w')
@@ -97,14 +107,20 @@ def run_my_code(input_text, language):
             output_text=i.split("\t")[2]
             break
 
-    #os.system(f"rm {data_root}data/tst-COMMON/wav/test.wav")
+    output_audio = "output_audio.mp3"
+    tts = gTTS(text=output_text, lang=lang)
+    tts.save(output_audio) 
+
+
+
+    #os.system(f"rm test.wav")
     f = open('input.txt', 'w')
     f.write("")
 
     f = open('input.txt', 'r')
     content = f. read()
     print(content)
-    return output_text
+    return output_text, output_audio
 
 install_fairseq()
 
@@ -123,7 +139,7 @@ output_textbox = gr.outputs.Textbox(label="Translated Text")
 iface = gr.Interface(
         fn=run_my_code, 
         inputs=[gr.inputs.Audio(source="microphone", type="filepath", label="Record something (in American English accent)"), gr.inputs.Radio(["Hindi", "Gujrati", "Bengali", "Tamil", "Nepali", "Marathi"], label="Language")], 
-        outputs=output_textbox,
+        outputs=[output_textbox, gr.outputs.Audio(label="Output speech", type="filepath", live=True)],
         title="English to Indic Language Translator")
 
 # Launch the interface