Add Redux and Podcast API

Author: LERM0

.gitignore

@@ -38,4 +38,5 @@ yarn-error.log*
 *.pem
 .DS_Store
 
-__pycache__
+__pycache__
+tmp

LOG.md

@@ -0,0 +1,6 @@
+
+Run with docker
+```sh
+docker build -t lerm0/core  . --platform=linux/amd64
+docker run -it -v /Users/LermoAI/apps/api/core-api:/app lerm0/core bash
+```

apps/api/core-api/Dockerfile

@@ -24,7 +24,9 @@ COPY . .
 RUN mkdir /.local && chmod 777 /.local
 RUN mkdir /.cache && chmod 777 /.cache
 RUN pip install -r requirements.txt
+RUN pip install git+https://github.com/myshell-ai/MeloTTS.git
+RUN python -m unidic download
 
 EXPOSE 8000
 
-ENTRYPOINT python main.py
+# ENTRYPOINT python main.py

apps/api/core-api/app/api/main.py

@@ -0,0 +1,24 @@
+from fastapi import APIRouter, HTTPException, Depends, Header, Request
+from fastapi.responses import StreamingResponse, FileResponse
+
+from app.engine import create_podcast_script, text_to_voice
+
+router = APIRouter()
+
+@router.post("/podcast")
+async def chat_with_model(request: Request):
+  try:
+    params = await request.json()
+    prompt = params.get("prompt")
+    content = params.get("content")
+    text = create_podcast_script(content)
+
+    prompt_array = prompt.split(", ")[:2] 
+    file_name = prompt_array[0]
+    speaker = prompt_array[1]
+    file_path_full = f"/tmp/voice/{file_name}.wav"
+
+    text_to_voice(text, file_path_full, speaker)
+    return FileResponse(path=file_path_full, media_type="audio/wav", filename=f"{file_name}.wav")
+  except Exception as e:
+    raise HTTPException(status_code=500, detail=str(e))

apps/api/core-api/app/engine.py

@@ -0,0 +1,27 @@
+from langchain_openai import ChatOpenAI
+from melo.api import TTS
+
+
+llm = None
+VOICE_ENGINE = None
+device = 'cpu'
+def init_engine():
+  global llm
+  global VOICE_ENGINE
+  if not llm:
+    llm = ChatOpenAI(model="gpt4", temperature=0)
+  if not VOICE_ENGINE:
+      VOICE_ENGINE = TTS(language='EN', device=device)
+
+def create_podcast_script(content: str):
+  prompt = """
+  Go deep into {content} and explain like human talk max 300 words.
+  """
+  result = llm.invoke(prompt.format(content=content))
+  return result.content
+
+def text_to_voice(text: str, file_path: str, speaker: str):
+    speed = 1
+    speaker_ids = VOICE_ENGINE.hps.data.spk2id
+    VOICE_ENGINE.tts_to_file(text, speaker_ids[speaker], file_path, speed=speed)
+    print("Voice Generated")

apps/api/core-api/app/main.py

@@ -1,6 +1,7 @@
 from fastapi import FastAPI
 from starlette.middleware.cors import CORSMiddleware
-from app.api import ping
+from app.api import ping, main
+from app.engine import init_engine
 
 app = FastAPI()
 
@@ -19,10 +20,12 @@
 
 @app.on_event("startup")
 async def startup():
+    init_engine()
     print("Starting up...")
 
 @app.on_event("shutdown")
 async def shutdown():
     print("Shutting down...")
 
-app.include_router(ping.router)
+app.include_router(ping.router)
+app.include_router(main.router)

apps/api/core-api/app/template.py

@@ -1,2 +1,4 @@
 fastapi==0.111.0
-uvicorn==0.29.0
+uvicorn==0.29.0
+langchain==0.2.1
+langchain_openai==0.1.7

apps/api/core-api/requirements.txt

@@ -2,7 +2,14 @@
 const nextConfig = {
   images: {
     domains: ["localhost"],
-  }
+  },
+  transpilePackages: [
+    '@ant-design',
+    "rc-util",
+    "rc-pagination",
+    "rc-picker",
+    "rc-tree",
+    "rc-table",]
 };
 
 export default nextConfig;