fix api keys

software/source/clients/base_device.py

@@ -152,6 +152,15 @@ class Device:
 
     async def play_audiosegments(self):
         """Plays them sequentially."""
+
+        mpv_command = ["mpv", "--no-cache", "--no-terminal", "--", "fd://0"]
+        mpv_process = subprocess.Popen(
+            mpv_command,
+            stdin=subprocess.PIPE,
+            stdout=subprocess.DEVNULL,
+            stderr=subprocess.DEVNULL,
+        )
+
         while True:
             try:
                 audio = await self.audiosegments.get()
@@ -161,6 +170,10 @@ class Device:
                     print(f"Time from request to playback: {elapsed_time} seconds")
                     self.playback_latency = None
 
+                if audio is not None:
+                    mpv_process.stdin.write(audio)  # type: ignore
+                    mpv_process.stdin.flush()  # type: ignore
+                """
                 args = ["ffplay", "-autoexit", "-", "-nodisp"]
                 proc = subprocess.Popen(
                     args=args,
@@ -171,7 +184,8 @@ class Device:
                 out, err = proc.communicate(input=audio)
                 proc.poll()
 
-                # play(audio)
+                play(audio)
+                """
                 # self.audiosegments.remove(audio)
                 # await asyncio.sleep(0.1)
             except asyncio.exceptions.CancelledError:
@@ -391,7 +405,7 @@ class Device:
                         # signed 16-bit little-endian format
                         sample_width=2,
                         # 24,000 Hz frame rate
-                        frame_rate=24000,
+                        frame_rate=16000,
                         # mono sound
                         channels=1,
                     )
@@ -400,6 +414,8 @@ class Device:
                     # print("audio segment was created")
                     await self.audiosegments.put(audio_bytes)
 
+                    # await self.audiosegments.put(audio)
+
                 # Run the code if that's the client's job
                 if os.getenv("CODE_RUNNER") == "client":
                     if message["type"] == "code" and "end" in message:
@@ -434,7 +450,8 @@ class Device:
     async def start_async(self):
         print("start async was called!!!!!")
         # Configuration for WebSocket
-        WS_URL = f"ws://{self.server_url}/ws"
+        WS_URL = f"ws://{self.server_url}"
+
         # Start the WebSocket communication
         asyncio.create_task(self.websocket_communication(WS_URL))
 

software/source/server/async_interpreter.py

@@ -24,6 +24,7 @@ from RealtimeSTT import AudioToTextRecorder
 import time
 import asyncio
 import json
+import os
 
 
 class AsyncInterpreter:
@@ -47,9 +48,7 @@ class AsyncInterpreter:
         elif self.interpreter.tts == "gtts":
             engine = GTTSEngine()
         elif self.interpreter.tts == "elevenlabs":
-            engine = ElevenlabsEngine(
-                api_key="sk_077cb1cabdf67e62b85f8782e66e5d8e11f78b450c7ce171"
-            )
+            engine = ElevenlabsEngine(api_key=os.environ["ELEVEN_LABS_API_KEY"])
         elif self.interpreter.tts == "system":
             engine = SystemEngine()
         else:
@@ -96,12 +95,12 @@ class AsyncInterpreter:
                 pass
 
             if "start" in chunk:
-                print("input received")
+                # print("Starting STT")
                 self.stt.start()
                 self._last_lmc_start_flag = time.time()
                 # self.interpreter.computer.terminal.stop() # Stop any code execution... maybe we should make interpreter.stop()?
             elif "end" in chunk:
-                print("running oi on input now")
+                # print("Running OI on input")
                 asyncio.create_task(self.run())
             else:
                 await self._add_to_queue(self._input_queue, chunk)
@@ -139,6 +138,7 @@ class AsyncInterpreter:
         print("STT LATENCY", self.stt_latency)
 
         # print(message)
+        end_interpreter = 0
 
         # print(message)
         def generate(message):
@@ -165,7 +165,7 @@ class AsyncInterpreter:
 
                         # Experimental: The AI voice sounds better with replacements like these, but it should happen at the TTS layer
                         # content = content.replace(". ", ". ... ").replace(", ", ", ... ").replace("!", "! ... ").replace("?", "? ... ")
-                        print("yielding this", content)
+                        # print("yielding this", content)
                         yield content
 
                 # Handle code blocks
@@ -214,6 +214,7 @@ class AsyncInterpreter:
         while True:
             if self.tts.is_playing():
                 start_tts = time.time()
+
                 break
             await asyncio.sleep(0.1)
         while True:
@@ -231,6 +232,7 @@ class AsyncInterpreter:
                 end_tts = time.time()
                 self.tts_latency = end_tts - start_tts
                 print("TTS LATENCY", self.tts_latency)
+                self.tts.stop()
                 break
 
     async def _on_tts_chunk_async(self, chunk):

software/source/server/async_server.py

@@ -2,6 +2,7 @@ import asyncio
 import traceback
 import json
 from fastapi import FastAPI, WebSocket, Header
+from fastapi.responses import PlainTextResponse
 from uvicorn import Config, Server
 from interpreter import interpreter as base_interpreter
 from .async_interpreter import AsyncInterpreter
@@ -18,38 +19,25 @@ base_interpreter.system_message = (
     "You are a helpful assistant that can answer questions and help with tasks."
 )
 base_interpreter.computer.import_computer_api = False
-base_interpreter.llm.model = "groq/mixtral-8x7b-32768"
-base_interpreter.llm.api_key = (
-    "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"
-)
+base_interpreter.llm.model = "groq/llama3-8b-8192"
+base_interpreter.llm.api_key = os.environ["GROQ_API_KEY"]
+print(base_interpreter.llm.api_key)
 base_interpreter.llm.supports_functions = False
+base_interpreter.auto_run = True
 
 os.environ["STT_RUNNER"] = "server"
 os.environ["TTS_RUNNER"] = "server"
 
 # Parse command line arguments for port number
+"""
 parser = argparse.ArgumentParser(description="FastAPI server.")
 parser.add_argument("--port", type=int, default=8000, help="Port to run on.")
 args = parser.parse_args()
-
+"""
 base_interpreter.tts = "elevenlabs"
 
 
-async def main():
-    """
-    sentry_sdk.init(
-        dsn="https://a1465f62a31c7dfb23e1616da86341e9@o4506046614667264.ingest.us.sentry.io/4507374662385664",
-        enable_tracing=True,
-        # Set traces_sample_rate to 1.0 to capture 100%
-        # of transactions for performance monitoring.
-        traces_sample_rate=1.0,
-        # Set profiles_sample_rate to 1.0 to profile 100%
-        # of sampled transactions.
-        # We recommend adjusting this value in production.
-        profiles_sample_rate=1.0,
-    )
-    """
-
+async def main(server_host, server_port):
     interpreter = AsyncInterpreter(base_interpreter)
 
     app = FastAPI()
@@ -62,6 +50,10 @@ async def main():
         allow_headers=["*"],  # Allow all headers
     )
 
+    @app.get("/ping")
+    async def ping():
+        return PlainTextResponse("pong")
+
     @app.post("/load_chat")
     async def load_chat(messages: List[Dict[str, Any]]):
         interpreter.interpreter.messages = messages
@@ -69,7 +61,7 @@ async def main():
         print("🪼🪼🪼🪼🪼🪼 Messages loaded: ", interpreter.active_chat_messages)
         return {"status": "success"}
 
-    @app.websocket("/ws")
+    @app.websocket("/")
     async def websocket_endpoint(websocket: WebSocket):
         await websocket.accept()
         try:
@@ -85,7 +77,7 @@ async def main():
                         await interpreter.input(data)
                     elif "bytes" in data:
                         await interpreter.input(data["bytes"])
-                        # print("SERVER FEEDING AUDIO")
+                        # print("RECEIVED INPUT", data)
                     elif "text" in data:
                         # print("RECEIVED INPUT", data)
                         await interpreter.input(data["text"])
@@ -111,7 +103,8 @@ async def main():
             if not websocket.client_state == "DISCONNECTED":
                 await websocket.close()
 
-    config = Config(app, host="0.0.0.0", port=8000, lifespan="on")
+    print(f"Starting server on {server_host}:{server_port}")
+    config = Config(app, host=server_host, port=server_port, lifespan="on")
     server = Server(config)
     await server.serve()
 

software/start.py

@@ -22,7 +22,7 @@ def run(
         help="Specify the server host where the server will deploy",
     ),
     server_port: int = typer.Option(
-        8000,
+        10001,
         "--server-port",
         help="Specify the server port where the server will deploy",
     ),
@@ -152,8 +152,8 @@ def _run(
             target=loop.run_until_complete,
             args=(
                 main(
-                    # server_host,
-                    # server_port,
+                    server_host,
+                    server_port,
                     # llm_service,
                     # model,
                     # llm_supports_vision,
@@ -196,7 +196,7 @@ def _run(
         module = importlib.import_module(
             f".clients.{client_type}.device", package="source"
         )
-        server_url = "0.0.0.0:8000"
+        # server_url = "0.0.0.0:8000"
         client_thread = threading.Thread(target=module.main, args=[server_url])
         print("client thread started")
         client_thread.start()