383 lines
18 KiB
Python
383 lines
18 KiB
Python
#!/usr/bin/env python3
|
|
"""AVC optometry phone agent — the Pipecat pipeline for a single inbound call.
|
|
|
|
Same VAD -> STT -> LLM -> TTS loop as pipecat-run/bot.py, but the ends are swapped
|
|
for telephony: audio arrives/leaves as 8 kHz mu-law over a Twilio Media Stream
|
|
(WebSocket), decoded by TwilioFrameSerializer. STT runs on the GPU; the LLM is the
|
|
local `activeblue-avc` fine-tune via Ollama; TTS is local Kokoro.
|
|
|
|
This module just builds + runs the pipeline for one connected call. server.py owns
|
|
the FastAPI/TwiML/WebSocket side and calls run_call() once per call.
|
|
"""
|
|
|
|
import os
|
|
|
|
import re
|
|
import time
|
|
|
|
from loguru import logger
|
|
|
|
from pipecat.adapters.schemas.function_schema import FunctionSchema
|
|
from pipecat.adapters.schemas.tools_schema import ToolsSchema
|
|
from pipecat.audio.vad.silero import SileroVADAnalyzer
|
|
from pipecat.audio.vad.vad_analyzer import VADParams
|
|
from pipecat.frames.frames import (
|
|
BotStoppedSpeakingFrame,
|
|
EndFrame,
|
|
EndTaskFrame,
|
|
Frame,
|
|
InputAudioRawFrame,
|
|
LLMFullResponseEndFrame,
|
|
LLMTextFrame,
|
|
TTSSpeakFrame,
|
|
)
|
|
from pipecat.pipeline.pipeline import Pipeline
|
|
from pipecat.pipeline.runner import PipelineRunner
|
|
from pipecat.pipeline.task import PipelineParams, PipelineTask
|
|
from pipecat.processors.aggregators.llm_context import LLMContext
|
|
from pipecat.processors.aggregators.llm_response_universal import LLMContextAggregatorPair
|
|
from pipecat.processors.audio.vad_processor import VADProcessor
|
|
from pipecat.processors.frame_processor import FrameDirection, FrameProcessor
|
|
from pipecat.serializers.twilio import TwilioFrameSerializer
|
|
from pipecat.services.anthropic.llm import AnthropicLLMService
|
|
from pipecat.services.deepgram.stt import DeepgramSTTService
|
|
from pipecat.services.kokoro.tts import KokoroTTSService
|
|
from pipecat.services.ollama.llm import OLLamaLLMService
|
|
from pipecat.transports.websocket.fastapi import (
|
|
FastAPIWebsocketParams,
|
|
FastAPIWebsocketTransport,
|
|
)
|
|
|
|
from practice import practice_summary
|
|
|
|
# ── Config (env-overridable) ─────────────────────────────────────────────────
|
|
HERE = os.path.dirname(os.path.abspath(__file__))
|
|
# Reuse the Kokoro model files already downloaded by the pipecat-run project.
|
|
MODEL_DIR = os.environ.get("KOKORO_MODEL_DIR", "/home/tocmo0nlord/pipecat-run/models")
|
|
|
|
OLLAMA_MODEL = os.environ.get("OLLAMA_MODEL", "activeblue-avc:latest")
|
|
OLLAMA_URL = os.environ.get("OLLAMA_URL", "http://127.0.0.1:11434/v1")
|
|
# Swappable LLM provider: "ollama" (local) or "anthropic" (Claude API). Same universal
|
|
# LLMContext drives both — only the service construction differs (see build_llm_service).
|
|
LLM_PROVIDER = os.environ.get("LLM_PROVIDER", "ollama").lower()
|
|
ANTHROPIC_API_KEY = os.environ.get("ANTHROPIC_API_KEY", "")
|
|
# Defaults to the most capable model. For low-latency PHONE voice, set ANTHROPIC_MODEL to
|
|
# claude-haiku-4-5 (fastest) or claude-sonnet-4-6 (balance) — see notes in build_llm_service.
|
|
ANTHROPIC_MODEL = os.environ.get("ANTHROPIC_MODEL", "claude-opus-4-8")
|
|
# In-call function-calling: AUTO by provider — ON for Claude (reliable tool calls → real-time
|
|
# Odoo booking), OFF for local Ollama (llama3.1:8b over-calls / leaks JSON). An explicit
|
|
# ENABLE_TOOLS env overrides the auto choice either way.
|
|
_enable_tools_env = os.environ.get("ENABLE_TOOLS")
|
|
ENABLE_TOOLS = (
|
|
_enable_tools_env.lower() in ("1", "true", "yes")
|
|
if _enable_tools_env is not None
|
|
else (LLM_PROVIDER == "anthropic")
|
|
)
|
|
LLM_TEMPERATURE = float(os.environ.get("LLM_TEMPERATURE", "0.3"))
|
|
LLM_MAX_TOKENS = int(os.environ.get("LLM_MAX_TOKENS", "160"))
|
|
KOKORO_VOICE = os.environ.get("KOKORO_VOICE", "af_heart")
|
|
# Real-time STT is Deepgram Nova-2: end-of-utterance events in <300ms (vs Whisper's
|
|
# 1-3s of chunk buffering, the main cause of non-reply / repeat-yourself). Whisper
|
|
# large-v3 is retained for post-call transcription only (Phase 3).
|
|
DEEPGRAM_API_KEY = os.environ.get("DEEPGRAM_API_KEY", "")
|
|
|
|
# Twilio sends 8 kHz mu-law on the wire — we run the PIPELINE at 16 kHz and let
|
|
# TwilioFrameSerializer resample to/from the 8 kHz wire. (Silero VAD, Deepgram, and
|
|
# Kokoro are all happy at 16 kHz.)
|
|
WIRE_SAMPLE_RATE = 8000 # Twilio mu-law on the wire (serializer handles this)
|
|
PIPELINE_SAMPLE_RATE = 16000 # internal rate Whisper/VAD actually need
|
|
|
|
# VAD tuning. Defaults (confidence 0.7 / min_volume 0.6) are desktop-mic values that can
|
|
# miss short/quiet 8 kHz telephony utterances like "yes" — loosen them for the phone.
|
|
VAD_CONFIDENCE = float(os.environ.get("VAD_CONFIDENCE", "0.5"))
|
|
VAD_MIN_VOLUME = float(os.environ.get("VAD_MIN_VOLUME", "0.3"))
|
|
VAD_START_SECS = float(os.environ.get("VAD_START_SECS", "0.2"))
|
|
VAD_STOP_SECS = float(os.environ.get("VAD_STOP_SECS", "0.5"))
|
|
|
|
# Agent persona name — purely for warmth; change/remove freely.
|
|
AGENT_NAME = os.environ.get("AGENT_NAME", "Sofia")
|
|
|
|
SYSTEM_PROMPT = (
|
|
f"You are {AGENT_NAME}, a warm, friendly receptionist for Advanced Vision Care, an "
|
|
"optometry practice with eight offices in South Florida. You are on a real phone call, so "
|
|
"talk like a helpful human being: natural, relaxed, and genuinely conversational — usually "
|
|
"just one short sentence at a time. Speak in English. Say numbers, dates, and times as "
|
|
"words a person would say.\n\n"
|
|
"Your job is to answer callers' questions and to take appointment requests. To book a "
|
|
"visit you need four things: which office or city, the reason for the visit, a preferred "
|
|
"day and time, and their name. Gather these naturally as the conversation flows — don't "
|
|
"interrogate, and never ask for something the caller already told you (people often give "
|
|
"their name or reason in their first sentence). You already have their number from caller "
|
|
"ID, so never ask for a phone number. When you have the details, repeat them back in one "
|
|
"warm sentence to confirm, and let them know a staff member will call to finalize the time.\n\n"
|
|
"Stay truthful and within your limits:\n"
|
|
"- Use ONLY the facts below for addresses, phone numbers, insurance, and services. Never "
|
|
"make any of these up.\n"
|
|
"- To find the right office, ask what CITY or AREA is most convenient for the caller. Do "
|
|
"NOT suggest or name a specific office yourself — you don't know where they are. Only after "
|
|
"they tell you their area, name the matching office; and only list locations if they ask "
|
|
"what offices exist.\n"
|
|
"- You cannot see a calendar, so never say a time is open or available — take the time as "
|
|
"a request that staff will confirm.\n"
|
|
"- Insurance: only confirm a plan that is in the list below. For any plan that is not "
|
|
"listed (UnitedHealthcare, Aetna, Cigna, and so on), don't say yes or no — say our staff "
|
|
"will verify their coverage.\n"
|
|
"- Hours are not published — say they vary by office and staff will confirm; never give "
|
|
"specific hours.\n"
|
|
"- You don't give medical advice and can't transfer calls. If the caller mentions an eye "
|
|
"problem, just note it as the reason and say a staff member or doctor will follow up.\n"
|
|
"- If you're not sure you heard something, simply ask them to repeat it.\n"
|
|
"- When the caller is all set, give a brief, warm closing that ends with the word "
|
|
"'Goodbye' — that ends the call, so only say it when you truly mean to.\n\n"
|
|
"PRACTICE FACTS:\n" + practice_summary()
|
|
)
|
|
|
|
|
|
def _build_tools() -> ToolsSchema:
|
|
# Only the booking action is a tool. Practice facts already live in the system prompt,
|
|
# so no get_practice_info tool (avoids needless calls/latency). callback_number is NOT
|
|
# required — we have the caller-ID and inject it in the handler.
|
|
return ToolsSchema(
|
|
standard_tools=[
|
|
FunctionSchema(
|
|
name="record_appointment_request",
|
|
description=(
|
|
"Record the caller's appointment request once you have their name and at "
|
|
"least the office/city and reason. Call this when the caller wants to book "
|
|
"a visit; staff will call back to confirm the exact time."
|
|
),
|
|
properties={
|
|
"patient_name": {"type": "string", "description": "Caller's full name"},
|
|
"location": {"type": "string", "description": "Which office/city the caller wants, e.g. Hialeah, Kendall, Tamarac"},
|
|
"reason": {"type": "string", "description": "Reason for the visit, e.g. annual exam, broken glasses, eye pain"},
|
|
"preferred_time": {"type": "string", "description": "Preferred day/time in the caller's words, if given"},
|
|
},
|
|
required=["patient_name"],
|
|
),
|
|
]
|
|
)
|
|
|
|
|
|
class EndCallProcessor(FrameProcessor):
|
|
"""Lets Sofia hang up. MUST sit between the LLM and the TTS: there it sees her reply
|
|
text (LLMTextFrame, flowing downstream) AND the upstream copy of BotStoppedSpeakingFrame
|
|
the output transport emits. It accumulates each reply; if the finished reply contains a
|
|
closing ('goodbye'/'adiós'), it waits until she's done speaking, then pushes EndTaskFrame
|
|
upstream — the task ends and TwilioFrameSerializer (auto_hang_up) drops the call."""
|
|
|
|
_CLOSINGS = ("goodbye", "good-bye", "good bye", "adiós", "adios", "hasta luego")
|
|
|
|
def __init__(self):
|
|
super().__init__()
|
|
self._buf = ""
|
|
self._should_end = False
|
|
|
|
@classmethod
|
|
def _is_closing(cls, text: str) -> bool:
|
|
t = (text or "").lower()
|
|
return any(c in t for c in cls._CLOSINGS)
|
|
|
|
async def process_frame(self, frame: Frame, direction: FrameDirection):
|
|
await super().process_frame(frame, direction)
|
|
if isinstance(frame, LLMTextFrame):
|
|
self._buf += frame.text
|
|
elif isinstance(frame, LLMFullResponseEndFrame):
|
|
if self._is_closing(self._buf):
|
|
self._should_end = True
|
|
logger.info("Sofia signalled closing -- will hang up after she finishes speaking")
|
|
self._buf = ""
|
|
elif isinstance(frame, BotStoppedSpeakingFrame) and self._should_end:
|
|
self._should_end = False
|
|
logger.info("Sofia closed the call -- ending task / hanging up")
|
|
await self.push_frame(EndTaskFrame(), FrameDirection.UPSTREAM)
|
|
await self.push_frame(frame, direction)
|
|
|
|
|
|
class AudioHeartbeat(FrameProcessor):
|
|
"""Diagnostic: logs how many inbound audio frames arrive every ~5s. If this keeps
|
|
ticking but VAD never fires, the issue is VAD/threshold; if it drops to 0 after a
|
|
turn, inbound audio stalled at the transport. Cheap, leave it on while stabilizing."""
|
|
|
|
def __init__(self):
|
|
super().__init__()
|
|
self._n = 0
|
|
self._t = time.time()
|
|
|
|
async def process_frame(self, frame: Frame, direction: FrameDirection):
|
|
await super().process_frame(frame, direction)
|
|
if isinstance(frame, InputAudioRawFrame):
|
|
self._n += 1
|
|
now = time.time()
|
|
if now - self._t >= 5:
|
|
logger.info(f"[audio-in] {self._n} frames in last {now - self._t:.0f}s")
|
|
self._n = 0
|
|
self._t = now
|
|
await self.push_frame(frame, direction)
|
|
|
|
|
|
def build_llm_service():
|
|
"""Build the LLM service for the selected provider. The universal LLMContext +
|
|
aggregators work with either, so only this construction differs (true A/B swap)."""
|
|
if LLM_PROVIDER == "anthropic":
|
|
if not ANTHROPIC_API_KEY:
|
|
raise RuntimeError("LLM_PROVIDER=anthropic but ANTHROPIC_API_KEY is not set")
|
|
logger.info(f"LLM provider: anthropic ({ANTHROPIC_MODEL})")
|
|
# NOTE: Opus 4.8/4.7 reject temperature/top_p/top_k (HTTP 400), so we omit them —
|
|
# this keeps the default Opus model working. For low-latency phone voice, prefer
|
|
# claude-haiku-4-5 (fastest) or claude-sonnet-4-6 over Opus. enable_prompt_caching
|
|
# caches the system prompt + growing conversation prefix (helps multi-turn cost/latency).
|
|
return AnthropicLLMService(
|
|
api_key=ANTHROPIC_API_KEY,
|
|
settings=AnthropicLLMService.Settings(
|
|
model=ANTHROPIC_MODEL,
|
|
enable_prompt_caching=True,
|
|
max_tokens=LLM_MAX_TOKENS,
|
|
),
|
|
)
|
|
logger.info(f"LLM provider: ollama ({OLLAMA_MODEL})")
|
|
return OLLamaLLMService(
|
|
settings=OLLamaLLMService.Settings(
|
|
model=OLLAMA_MODEL,
|
|
temperature=LLM_TEMPERATURE,
|
|
max_tokens=LLM_MAX_TOKENS,
|
|
),
|
|
base_url=OLLAMA_URL,
|
|
)
|
|
|
|
|
|
async def run_agent(transport, caller_number=None, call_sid=None, do_capture=True):
|
|
"""Build + run the AVC voice agent on a given transport. Shared by the phone path
|
|
(Twilio Media Stream) and the browser path (WebRTC) — same prompt, model, voice, and
|
|
booking/hang-up logic; only the transport differs. do_capture writes the post-call
|
|
appointment to Odoo (on for phone; off for browser testing so it doesn't make cards)."""
|
|
stt = DeepgramSTTService(
|
|
api_key=DEEPGRAM_API_KEY,
|
|
settings=DeepgramSTTService.Settings(
|
|
model="nova-2",
|
|
language="en-US",
|
|
smart_format=True,
|
|
punctuate=True,
|
|
interim_results=False, # final transcripts only — avoids double-firing
|
|
utterance_end_ms=1000, # ms of silence before end-of-utterance fires
|
|
),
|
|
)
|
|
llm = build_llm_service()
|
|
# In-call booking tool — only registered when ENABLE_TOOLS is on (auto: Claude yes,
|
|
# local Ollama no, since llama3.1:8b over-calls/leaks). The handler is a closure so it
|
|
# can stamp the verified caller-ID + call_sid onto the lead (the model never supplies a
|
|
# phone number — we don't ask for one). With tools on, this writes the Odoo lead IN-CALL,
|
|
# so the post-call extraction is skipped below to avoid a duplicate.
|
|
if ENABLE_TOOLS:
|
|
async def _record_appointment(params):
|
|
args = params.arguments or {}
|
|
if do_capture:
|
|
from practice import persist_appointment
|
|
persist_appointment({
|
|
"call_sid": call_sid,
|
|
"patient_name": args.get("patient_name"),
|
|
"callback_number": caller_number, # verified caller-ID, not model-supplied
|
|
"location": args.get("location"),
|
|
"reason": args.get("reason"),
|
|
"preferred_time": args.get("preferred_time"),
|
|
"source": "in_call_tool",
|
|
})
|
|
else:
|
|
logger.info(f"[capture off] would record appointment: {args.get('patient_name')} / {args.get('location')}")
|
|
await params.result_callback(
|
|
{"status": "recorded", "message": "Recorded — staff will call to confirm the time."}
|
|
)
|
|
|
|
llm.register_function("record_appointment_request", _record_appointment)
|
|
|
|
tts = KokoroTTSService(
|
|
model_path=os.path.join(MODEL_DIR, "kokoro-v1.0.onnx"),
|
|
voices_path=os.path.join(MODEL_DIR, "voices-v1.0.bin"),
|
|
settings=KokoroTTSService.Settings(voice=KOKORO_VOICE),
|
|
)
|
|
vad = VADProcessor(vad_analyzer=SileroVADAnalyzer(params=VADParams(
|
|
confidence=VAD_CONFIDENCE,
|
|
start_secs=VAD_START_SECS,
|
|
stop_secs=VAD_STOP_SECS,
|
|
min_volume=VAD_MIN_VOLUME,
|
|
)))
|
|
heartbeat = AudioHeartbeat()
|
|
|
|
context_kwargs = {"messages": [{"role": "system", "content": SYSTEM_PROMPT}]}
|
|
if ENABLE_TOOLS:
|
|
context_kwargs["tools"] = _build_tools()
|
|
context = LLMContext(**context_kwargs)
|
|
agg = LLMContextAggregatorPair(context)
|
|
endcall = EndCallProcessor()
|
|
|
|
pipeline = Pipeline(
|
|
[
|
|
transport.input(),
|
|
heartbeat,
|
|
vad,
|
|
stt,
|
|
agg.user(),
|
|
llm,
|
|
endcall,
|
|
tts,
|
|
transport.output(),
|
|
agg.assistant(),
|
|
]
|
|
)
|
|
|
|
task = PipelineTask(
|
|
pipeline,
|
|
params=PipelineParams(
|
|
audio_in_sample_rate=PIPELINE_SAMPLE_RATE,
|
|
audio_out_sample_rate=PIPELINE_SAMPLE_RATE,
|
|
allow_interruptions=True,
|
|
),
|
|
)
|
|
|
|
@transport.event_handler("on_client_connected")
|
|
async def on_client_connected(transport, client):
|
|
logger.info("Client connected -- greeting")
|
|
await task.queue_frames(
|
|
[TTSSpeakFrame(
|
|
f"Thank you for calling Advanced Vision Care, this is {AGENT_NAME}. "
|
|
"How can I help you today?"
|
|
)]
|
|
)
|
|
|
|
@transport.event_handler("on_client_disconnected")
|
|
async def on_client_disconnected(transport, client):
|
|
logger.info("Client disconnected -- ending task")
|
|
await task.queue_frame(EndFrame())
|
|
|
|
runner = PipelineRunner(handle_sigint=False)
|
|
await runner.run(task)
|
|
|
|
# Call is over. Post-call extraction is the capture path ONLY when in-call tools are
|
|
# off (local Ollama). With tools on (Claude), the booking was already written in-call,
|
|
# so skip extraction to avoid a duplicate lead.
|
|
if do_capture and not ENABLE_TOOLS:
|
|
try:
|
|
from extract import extract_and_record
|
|
|
|
await extract_and_record(
|
|
context.messages, OLLAMA_URL, OLLAMA_MODEL,
|
|
call_sid=call_sid, caller_number=caller_number,
|
|
)
|
|
except Exception:
|
|
logger.exception("Post-call appointment extraction failed")
|
|
|
|
|
|
async def run_call(websocket, serializer: TwilioFrameSerializer, caller_number=None, call_sid=None):
|
|
"""Phone entrypoint: wrap the Twilio Media Stream in a transport, run the shared agent."""
|
|
transport = FastAPIWebsocketTransport(
|
|
websocket=websocket,
|
|
params=FastAPIWebsocketParams(
|
|
audio_in_enabled=True,
|
|
audio_out_enabled=True,
|
|
audio_in_sample_rate=PIPELINE_SAMPLE_RATE,
|
|
audio_out_sample_rate=PIPELINE_SAMPLE_RATE,
|
|
add_wav_header=False,
|
|
serializer=serializer,
|
|
),
|
|
)
|
|
await run_agent(transport, caller_number=caller_number, call_sid=call_sid, do_capture=True)
|