feat: add voice transcription

app/api/v1/voice.py

@@ -1,6 +1,6 @@
 from fastapi import APIRouter
-from app.models.voice import VoiceRequest, VoiceResponse
-from app.services.voice import generate_voice
+from app.models.voice import VoiceRequest, VoiceResponse, TranscriptionRequest, TranscriptionResponse
+from app.services.voice import generate_voice, generate_transcription
 
 router = APIRouter()
 
@@ -8,3 +8,9 @@ router = APIRouter()
 async def voice(request: VoiceRequest):
     voice = await generate_voice(request.text)
     return VoiceResponse(voice=voice)
+
+@router.post("/transcription", response_model=TranscriptionResponse)
+async def transcription(request: TranscriptionRequest):
+    transcription = await generate_transcription(request.audio_file_path)
+    return TranscriptionResponse(transcription=transcription)
+

app/core/openai_voice_transcription_client.py

@@ -20,8 +20,36 @@ async def generate_voice(messages: list):
     try:
         response = client.chat.completions.create(
             model=OPENAI_AUDIO_MODEL,
-            response_format="mp3",
             messages=messages,
+            max_tokens=1000,
+            temperature=0.7,
+            stream=False
+        )
+        
+        if not response.choices or not response.choices[0].message.content:
+            return "No response content from the model"
+            
+        return response.choices[0].message.content
+        
+    except OpenAIError as e:
+        error_msg = f"OpenAI API Error: {str(e)}"
+        print(error_msg)
+        raise Exception(error_msg) from e
+    except Exception as e:
+        error_msg = f"Unexpected error: {str(e)}"
+        print(error_msg)
+        raise Exception(error_msg) from e
+    
+async def generate_transcription(audio_file_path: str) -> str:
+    if not audio_file_path:
+        raise ValueError("Audio file path cannot be empty")
+
+    try:
+        response = client.audio.transcriptions.create(
+            model=OPENAI_AUDIO_MODEL,
+            file=audio_file_path,
+            response_format="text",
+            language="id"
         )
         
         if not response.choices or not response.choices[0].message.content:

app/models/voice.py

@@ -1,8 +1,17 @@
 from pydantic import BaseModel
+from typing import Optional
 
-
+# Text-to-Speech Models
 class VoiceRequest(BaseModel):
     text: str
 
 class VoiceResponse(BaseModel):
-    voice: str
+    voice_output: str
+
+# Speech-to-Text Models
+class TranscriptionRequest(BaseModel):
+    audio_file_path: str
+    target_language: Optional[str] = "id"  # Default to English
+
+class TranscriptionResponse(BaseModel):
+    text: str

app/services/voice.py

@@ -1,7 +1,10 @@
-from app.core.openai_voice_client import generate_voice
+from app.core.openai_voice_transcription_client import generate_voice, generate_transcription
 
 async def generate_voice(text: str) -> str:
     voice = await generate_voice(text)
     return voice
 
+async def generate_transcription(audio_file_path: str) -> str:
+    transcription = await generate_transcription(audio_file_path)
+    return transcription