Update app.py

app.py

@@ -4,6 +4,7 @@ import whisper
 from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
 from gtts import gTTS
 import IPython.display as ipd
+import numpy as np
 
 # Load Whisper STT model
 whisper_model = whisper.load_model("base")
@@ -12,9 +13,14 @@ whisper_model = whisper.load_model("base")
 tokenizer = AutoTokenizer.from_pretrained("alirezamsh/small100")
 model = AutoModelForSeq2SeqLM.from_pretrained("alirezamsh/small100")
 
-def translate_speech(audio_file, target_lang):
+def translate_speech(audio, target_lang):
+    # Save audio as a temporary file
+    audio_path = "recorded_audio.wav"
+    with open(audio_path, "wb") as f:
+        f.write(audio)
+
     # Load audio
-    audio = whisper.load_audio(audio_file)
+    audio = whisper.load_audio(audio_path)
     audio = whisper.pad_or_trim(audio)
     mel = whisper.log_mel_spectrogram(audio).to(whisper_model.device)
 
@@ -41,18 +47,15 @@ def translate_speech(audio_file, target_lang):
     return audio_path
 
 def translate_speech_interface(audio, target_lang):
-    audio_path = "recorded_audio.wav"
-    with open(audio_path, "wb") as f:
-        f.write(audio.read())
-
-    translated_audio = translate_speech(audio_path, target_lang)
+    translated_audio = translate_speech(audio, target_lang)
     translated_audio = open(translated_audio, "rb")
 
     return translated_audio
 
 # Define the Gradio interface
-audio_recording = gr.inputs.Audio(source="microphone", type="wav", label="Record your speech")
+audio_recording = gr.inputs.Audio(source="microphone", type="numpy", label="Record your speech")
 target_language = gr.inputs.Dropdown(["en", "ru", "fr"], label="Target Language")
 output_audio = gr.outputs.Audio(type="audio/mpeg", label="Translated Audio")
 
 gr.Interface(fn=translate_speech_interface, inputs=[audio_recording, target_language], outputs=output_audio, title="Speech Translator").launch()
+