added noise cancellation

app.py

@@ -78,8 +78,29 @@ language_info={
     'Other': 'Other'
 }
         
+import librosa
+import noisereduce as nr
+from pydub import AudioSegment
+
+def enhance_and_denoise_audio(file_path):
+    # Load the audio file
+    data, rate = librosa.load(file_path, sr=None)
+
+    # Perform noise reduction
+    reduced_noise_audio = nr.reduce_noise(y=data, sr=rate)
+
+    # Increase the volume by 10 dB (using PyDub)
+    audio_segment = AudioSegment(reduced_noise_audio.tobytes(), frame_rate=rate, sample_width=reduced_noise_audio.dtype.itemsize, channels=1)
+    enhanced_audio = audio_segment + 10
+
+    # Export the enhanced audio to the same file
+    enhanced_audio.export(file_path, format="wav")
+
+    return file_path        
+
 def translate(audio_file,lan):
     message=""
+    audio_file=enhance_and_denoise_audio(audio_file)
     with open(audio_file, 'rb') as f:
         result = openai.Audio.translate("whisper-1", f)
         text=result.text
@@ -96,6 +117,7 @@ def translate(audio_file,lan):
 
 def transcription(audio_file,input_lang):
     global language_info
+    audio_file=enhance_and_denoise_audio(audio_file)
     with open(audio_file, 'rb') as f:
         if input_lang=="Other":
             result = openai.Audio.transcribe("whisper-1", f)
@@ -104,7 +126,9 @@ def transcription(audio_file,input_lang):
             result = openai.Audio.transcribe("whisper-1", f,language=language_info[input_lang])
             th.current_thread().return_value=result.text
 
-    
+
+
+
 
 def func(audio_file,input_lang,lan,state="",state1=""):
     
@@ -127,8 +151,7 @@ def func(audio_file,input_lang,lan,state="",state1=""):
     return state,state1
 
 
-def gpt_api(text):
-    
+def gpt_api(text,language):
     if text=="":
         return ""
     if len(text)>2000:
@@ -136,7 +159,7 @@ def gpt_api(text):
     completion = openai.ChatCompletion.create(
     model="gpt-3.5-turbo",
     messages=[
-    {"role": "system", "content": "your task is to make a concise summery of text in the same language used in given text"},
+    {"role": "system", "content": f"your task is to make a concise summery and useful summery from the given text in {language}."},
     {"role": "user", "content":text}
 ]
 )
@@ -145,9 +168,9 @@ def gpt_api(text):
     th.current_thread().return_value=message
 
 
-def make_summery(text,text1):
-    t1=th.Thread(target=gpt_api, args=(text,))
-    t2=th.Thread(target=gpt_api, args=(text,))
+def make_summery(text,text1,input_lan,output_lan):
+    t1=th.Thread(target=gpt_api, args=(text,input_lan,))
+    t2=th.Thread(target=gpt_api, args=(text1,output_lan,))
     t1.start()
     t2.start()
     t1.join()
@@ -203,7 +226,7 @@ with gr.Blocks(theme=gr.themes.Soft(),css=css) as app:
         sumer_tr=gr.Textbox(label="Summery of Translation",interactive=False,lines=4,max_lines=4)
         
         # pass
-        summery.click(make_summery,[text,text1],[sumer_ts,sumer_tr],cancels=[st],queue=False)
+        summery.click(make_summery,[text,text1,input_lan,lan],[sumer_ts,sumer_tr],cancels=[st],queue=False)
         clear_output.click(clear_output_data,[],[text,text1,sumer_tr,sumer_ts],cancels=[st],queue=False)
         # gr.update(visible=True)
 

requirements.txt

@@ -70,3 +70,6 @@ urllib3==2.0.7
 uvicorn==0.24.0.post1
 websockets==11.0.3
 yarl==1.9.2
+noisereduce==3.0.0
+librosa==0.10.1
+pydub==0.25.1