65 lines
1.7 KiB
Python
65 lines
1.7 KiB
Python
from openai import OpenAI
|
|
from fastapi import FastAPI, File, Response, Request
|
|
from fastapi.responses import FileResponse
|
|
from pydantic import BaseModel
|
|
import whisper
|
|
|
|
|
|
app = FastAPI()
|
|
openAI_clinet = OpenAI()
|
|
model = whisper.load_model("base")
|
|
|
|
|
|
class ConversationMessege(BaseModel):
|
|
role: str
|
|
content: str
|
|
|
|
|
|
class Conversation(BaseModel):
|
|
messages: list[ConversationMessege]
|
|
|
|
|
|
@app.post("/get-text")
|
|
def get_text(response: Response, audio: bytes = File()):
|
|
response.headers["Access-Control-Allow-Origin"] = "*"
|
|
with open("audio", "wb") as f:
|
|
f.write(audio)
|
|
# transcript = openAI_clinet.audio.transcriptions.create(
|
|
# model="whisper-1",
|
|
# file=audio,
|
|
# response_format="text",
|
|
# RequestBody
|
|
# )
|
|
result = model.transcribe("audio")
|
|
data = {"len": len(audio), "user-transcript": result["text"]}
|
|
return data
|
|
|
|
|
|
@app.post("/conversation")
|
|
async def get_next_response(request: Request, response: Response):
|
|
response.headers["Access-Control-Allow-Origin"] = "*"
|
|
# role = "test"
|
|
# res_msg = "temp test response"
|
|
messages = await request.json()
|
|
res = openAI_clinet.chat.completions.create(
|
|
model="gpt-3.5-turbo",
|
|
messages=messages,
|
|
)
|
|
res_msg = res.choices[0].message.content
|
|
role = res.choices[0].message.role
|
|
print(messages)
|
|
print(res_msg)
|
|
return {"role": role, "content": res_msg}
|
|
|
|
|
|
@app.post("/speak", response_class=FileResponse)
|
|
def tts(text: str, response: Response):
|
|
response.headers["Access-Control-Allow-Origin"] = "*"
|
|
res = openAI_clinet.audio.speech.create(
|
|
model="tts-1",
|
|
voice="nova",
|
|
input=text,
|
|
)
|
|
res.stream_to_file("tts.mp3")
|
|
return "tts.mp3"
|