Update app.py

app.py

@@ -2,24 +2,31 @@ import os
 import subprocess
 import openai
 import gradio as gr
+import requests
+from gtts import gTTS
 
 
-import os
-
 openai.api_key = os.getenv("OPENAI_API_KEY")
 
+API_URL = "https://api-inference.huggingface.co/models/lyimo/whisper-small-sw2"
+headers = {"Authorization": f"Bearer {os.getenv('HUGGINGFACE_API_KEY')}"}
 
-def transcribe(audio):
-   with open(audio, "rb") as audio_file:
-       transcript = openai.Audio.transcribe("whisper-1", audio_file)
-   return transcript["text"]
+def query(filename):
+    with open(filename, "rb") as f:
+        data = f.read()
+    response = requests.post(API_URL, headers=headers, data=data)
+    return response.json()
 
 
+def transcribe(audio):
+    output = query(audio)
+    return output["text"]
+
 def generate_response(transcribed_text):
    response = openai.ChatCompletion.create(
        model="gpt-3.5-turbo",
        messages=[
-           {"role": "system", "content": " All your answers should be in swahili only, users undertands swahili only so here we start... Wewe ni mtaalamu wa viazi lishe na utajibu maswali yote kwa kiswahili tu!"},
+           {"role": "system", "content": " All your answers should be in swahili only, users undertands swahili only, so here we start... Wewe ni mtaalamu wa viazi lishe na utajibu maswali yote kwa kiswahili tu!"},
            {"role": "user", "content": "Mambo vipi?"},
            {"role": "assistant", "content": """Salama je una swali lolote kuhusu viazi lishe?"""},
            {"role": "user", "content": "nini maana ya Viazi lishe?"},
@@ -37,32 +44,28 @@ def generate_response(transcribed_text):
    )
    return response['choices'][0]['message']['content']
 
-
 def inference(text):
-   output_file = "tts_output.wav"
-   cmd = ['tts', '--text', text, '--out_path', output_file]
-   subprocess.run(cmd, check=True)
-   return output_file
-
+    output_file = "tts_output.wav"
+    tts = gTTS(text, lang="sw")
+    tts.save(output_file)
+    return output_file
 
 def process_audio_and_respond(audio):
-   text = transcribe(audio)
-   response_text = generate_response(text)
-   output_file = inference(response_text)
-   return response_text, output_file
-
+    text = transcribe(audio)
+    response_text = generate_response(text)
+    output_file = inference(response_text)
+    return response_text, output_file
 
 demo = gr.Interface(
-   process_audio_and_respond,
-   gr.inputs.Audio(source="microphone", type="filepath", label="Bonyeza kitufe cha kurekodi na uliza swali lako"),
-   [gr.outputs.Textbox(label="Jibu (kwa njia ya maandishi)"), gr.outputs.Audio(type="filepath", label="Jibu kwa njia ya sauti (Bofya kusikiliza Jibu)")],
-   title="Mtaalamu wa Viazi Lishe",
-   description="Uliza Mtaalamu wetu swali lolote Kuhusu viazi Lishe",
-   theme="compact",
-   layout="vertical",
-   allow_flagging=False,
-   live=True,
+    process_audio_and_respond,
+    gr.inputs.Audio(source="microphone", type="filepath", label="Bonyeza kitufe cha kurekodi na uliza swali lako"),
+    [gr.outputs.Textbox(label="Jibu (kwa njia ya maandishi)"), gr.outputs.Audio(type="filepath", label="Jibu kwa njia ya sauti (Bofya kusikiliza Jibu)")],
+    title="Mtaalamu wa Viazi Lishe",
+    description="Uliza Mtaalamu wetu swali lolote Kuhusu viazi Lishe",
+    theme="compact",
+    layout="vertical",
+    allow_flagging=False,
+    live=True,
 )
 
-
-demo.launch()
+demo.launch()