Merge branch 'main' into p2p_ios_demo

This commit is contained in:
Filipi Fuchter
2025-04-07 16:50:58 -03:00
19 changed files with 6332 additions and 14 deletions

View File

@@ -9,24 +9,27 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
### Added
- Added new processors `ProducerProcessor` and `ConsumerProcessor`. The producer
processor processes frames from the pipeline and decides whether the consumers
should consume it or not. If so, the same frame that is received by the
producer is sent to the consumer. There can be multiple consumers per
- Added new processors `ProducerProcessor` and `ConsumerProcessor`. The
producer processor processes frames from the pipeline and decides whether the
consumers should consume it or not. If so, the same frame that is received by
the producer is sent to the consumer. There can be multiple consumers per
producer. These processors can be useful to push frames from one part of a
pipeline to a different one (e.g. when using `ParallelPipeline`).
- Improvements for the `SmallWebRTCTransport`:
- Wait until the pipeline is ready before triggering the `connected` event.
- Queue messages if the data channel is not ready.
- Update the aiortc dependency to fix an issue where the 'video/rtx' MIME type
was incorrectly handled as a codec retransmission.
- Update the aiortc dependency to fix an issue where the 'video/rtx' MIME
type was incorrectly handled as a codec retransmission.
- Avoid initial video delays.
### Fixed
- Fixed an issue in the Azure TTS services where the language was being set
incorrectly.
- Fixed `SmallWebRTCTransport` to support dynamic values for
`TransportParams.audio_out_10ms_chunks`. Previously, it only worked with 20ms
`TransportParams.audio_out_10ms_chunks`. Previously, it only worked with 20ms
chunks.
- Fixed an issue where `LLMAssistantContextAggregator` would prevent a

View File

@@ -0,0 +1,178 @@
# Handling PSTN/SIP Dial-in on Pipecat Cloud
This repository contains two server implementations for handling
the pinless dial-in workflow in Pipecat Cloud. This is the companion to the
Pipecat Cloud [pstn_sip starter image](https://github.com/daily-co/pipecat-cloud-images/tree/main/pipecat-starters/pstn_sip).
In addition you can use `/api/dial` to trigger dial-out, and
eventually, call-transfers.
1. [FastAPI Server](fastapi-webhook-server/README.md) -
A FastAPI implementation that handles PSTN (Public Switched Telephone
Network) and SIP (Session Initiation Protocol) calls using the Daily API.
2. [Next.js Serverless](nextjs-webhook-server/README.md) -
A Next.js API implementation designed for deployment on Vercel's
serverless platform.
Both implementations provide:
- HMAC signature validation for pinless webhook
- Structured logging
- Support for dial-in and dial-out settings
- Voicemail detection and call transfer functionality (coming soon)
- Test request handling
## Choosing an Implementation
- Use the **FastAPI Server** if you:
- Need a standalone server
- Prefer Python and FastAPI
- Want to deploy to traditional hosting platforms
- Use the **Next.js Serverless** implementation if you:
- Want serverless deployment
- Prefer JavaScript/TypeScript
- Already use Next.js and Vercel for other projects
- Need quick scaling and zero maintenance
## Prerequisites
### Environment Variables
Both implementations require similar environment variables:
- `PIPECAT_CLOUD_API_KEY`: Pipecat Cloud API Key, begins with pk\_\*
- `AGENT_NAME`: Your Daily agent name
- `PINLESS_HMAC_SECRET`: Your HMAC secret for request verification
- `LOG_LEVEL`: (Optional) Logging level (defaults to 'info')
See the individual README files in each implementation directory for
specific setup instructions.
### Phone number setup
You can buy a phone number through the Pipecat Cloud Dashboard:
1. Go to `Settings` > `Telephony`
2. Follow the UI to purchase a phone number
3. Configure the webhook URL to receive incoming calls (e.g. `https://my-webhook-url.com/api/dial`)
Or purchase the number using Daily's
[PhoneNumbers API](https://docs.daily.co/reference/rest-api/phone-numbers).
```bash
curl --request POST \
--url https://api.daily.co/v1/domain-dialin-config \
--header 'Authorization: Bearer $TOKEN' \
--header 'Content-Type: application/json' \
--data-raw '{
"type": "pinless_dialin",
"name_prefix": "Customer1",
"phone_number": "+1PURCHASED_NUM",
"room_creation_api": "https://example.com/api/dial",
"hold_music_url": "https://example.com/static/ringtone.mp3",
"timeout_config": {
"message": "No agent is available right now"
}
}'
```
The API will return a static SIP URI (`sip_uri`) that can be called
from other SIP services.
### `room_creation_api`
To make and receive calls currently you have to host a server that
handles incoming calls. In the coming weeks, incoming calls will be
directly handled within Daily and we will expose an endpoint similar
to `{service}/start` that will manage this for you.
In the meantime, the server described below serves as the webhook
handler for the `room_creation_api`. Configure your pinless phone
number or SIP interconnect to the `ngrok` tunnel or
the actual server URL, append `/api/dial` to the webhook URL.
## Example curl commands
Note: Replace `http://localhost:3000` with your actual server URL and
phone numbers with valid values for your use case.
### Dialin Request
The server will receive a request when a call is received from Daily.
### Dialout Request
Dial a number, will use any purchased number
```bash
curl -X POST http://localhost:3000/api/dial \
-H "Content-Type: application/json" \
-d '{
"dialout_settings": [
{
"phoneNumber": "+1234567890",
}
]
}'
```
Dial a number with callerId, which is the UUID of a purchased number.
```bash
curl -X POST http://localhost:3000/api/dial \
-H "Content-Type: application/json" \
-d '{
"dialout_settings": [
{
"phoneNumber": "+1234567890",
"callerId": "purchased_phone_uuid"
}
]
}'
```
Dial a number
```bash
curl -X POST http://localhost:3000/api/dial \
-H "Content-Type: application/json" \
-d '{
"dialout_settings": [
{
"phoneNumber": "+1234567890",
"callerId": "purchased_phone_uuid"
}
]
}'
```
### Advanced Request with Voicemail Detection
```bash
curl -X POST http://localhost:3000/api/dial \
-H "Content-Type: application/json" \
-d '{
"To": "+1234567890",
"From": "+1987654321",
"callId": "call-uuid-123",
"callDomain": "domain-uuid-456",
"dialout_settings": [
{
"phoneNumber": "+1234567890",
"callerId": "purchased_phone_uuid"
}
],
"voicemail_detection": {
"testInPrebuilt": true
},
"call_transfer": {
"mode": "dialout",
"speakSummary": true,
"storeSummary": true,
"operatorNumber": "+1234567890",
"testInPrebuilt": true
}
}'
```

View File

@@ -0,0 +1,98 @@
# FastAPI server for handling Daily PSTN/SIP Webhook
A FastAPI server that handles PSTN (Public Switched Telephone Network) and SIP (Session Initiation Protocol) calls using the Daily API.
## Setup
1. Clone the repository
2. Navigate to the `fastapi-webhook-server` directory:
```bash
cd fastapi-webhook-server
```
3. Install dependencies:
```bash
pip install -r requirements.txt
```
4. Copy `env.example` to `.env`:
```bash
cp env.example .env
```
5. Update `.env` with your credentials:
- `AGENT_NAME`: Your Daily agent name
- `PIPECAT_CLOUD_API_KEY`: Your Daily API key
- `PINLESS_HMAC_SECRET`: Your HMAC secret for request verification
## Running the Server
Start the server:
```bash
python server.py
```
The server will run on `http://localhost:7860` and you can expose it via ngrok for testing:
```bash
`ngrok http 7860`
```
> Tip: Use a subdomain for a consistent URL (e.g. `ngrok http -subdomain=mydomain http://localhost:7860`)
## API Endpoints
### GET /
Health check endpoint that returns a "Hello, World!" message.
### POST /api/dial
Initiates a PSTN/SIP call with the following request body format:
```json
{
"To": "+14152251493",
"From": "+14158483432",
"callId": "string-contains-uuid",
"callDomain": "string-contains-uuid",
"dialout_settings": [
{
"phoneNumber": "+14158483432",
"callerId": "+14152251493"
}
],
"voicemail_detection": {
"testInPrebuilt": true
},
"call_transfer": {
"mode": "dialout",
"speakSummary": true,
"storeSummary": true,
"operatorNumber": "+14152250006",
"testInPrebuilt": true
}
}
```
#### Response
Returns a JSON object containing:
- `status`: Success/failure status
- `data`: Response from Daily API
- `room_properties`: Properties of the created Daily room
## Error Handling
- 401: Invalid signature
- 400: Invalid authorization header (e.g. missing Daily API key in bot.py)
- 405: Method not allowed (e.g. incorrect route on the webhook URL)
- 500: Server errors (missing API key, network issues)
- Other status codes are passed through from the Daily API

View File

@@ -0,0 +1,3 @@
AGENT_NAME="your-agent-name"
PIPECAT_CLOUD_API_KEY="your-daily-api-key"
PINLESS_HMAC_SECRET="hmac-secret-pinless-dialin"

View File

@@ -0,0 +1,6 @@
fastapi
uvicorn
python-dotenv
requests
pydantic
loguru

View File

@@ -0,0 +1,201 @@
#
# Copyright (c) 2025, Daily
#
# SPDX-License-Identifier: BSD 2-Clause License
#
# server.py
import base64 # for calculating hmac signature
import hmac
import os # for accessing environment variables
import time # for setting expiration time
from typing import Any, Dict, List, Optional
import requests
from dotenv import load_dotenv
from fastapi import FastAPI, HTTPException, Request
from loguru import logger
from pydantic import BaseModel, Field
load_dotenv(override=True)
app = FastAPI()
class RoomRequest(BaseModel):
test: Optional[str] = Field(None, alias="Test", description="Test field")
To: Optional[str] = Field(None, alias="to", description="Destination phone number")
From: Optional[str] = Field(None, alias="from", description="Source phone number")
callId: Optional[str] = Field(None, alias="call_id", description="Unique call identifier")
callDomain: Optional[str] = Field(
None, alias="call_domain", description="Call domain identifier"
)
dialout_settings: Optional[List[Dict[str, Any]]] = Field(
None, description="An array of phone numbers or SIP URIs to dialout to"
)
voicemail_detection: Optional[Dict[str, Any]] = Field(
None, description="A flag to perform voicemail or answeing-machine detection"
)
call_transfer: Optional[Dict[str, Any]] = Field(None, description="to initiate a call transfer")
class Config:
populate_by_name = True
alias_generator = None
"""
body can contain any fields, but for handling PSTN/SIP,
we recommend sending the following custom values:
dialin, dialout, voicemail detection, and call transfer
"To": "+14152251493",
"From": "+14158483432",
"callId": "string-contains-uuid",
"callDomain": "string-contains-uuid"
These need to be remapped to dialin_settings
"dialout_settings": [
{"phoneNumber": "+14158483432", "callerId": "+14152251493"},
{"sipUri": "sip:username@sip.hostname"}
],
},
voicemail_detection:{
testInPrebuilt: true
},
"call_transfer": {
"mode": "dialout",
"speakSummary": true,
"storeSummary": true,
"operatorNumber": "+14152250006",
"testInPrebuilt": true
}
"""
@app.get("/")
async def read_root():
return {"message": "Hello, World!"}
@app.post("/api/dial")
async def dial(request: RoomRequest, raw_request: Request):
logger.info("Incoming request to /dial:")
logger.info(f"Headers: {dict(raw_request.headers)}")
raw_body = await raw_request.body()
raw_body_str = raw_body.decode()
logger.info(f"Raw body: {raw_body_str}")
logger.info(f"Parsed body: {request.dict()}")
# calculate signature and compare/verify
hmac_secret = os.getenv("PINLESS_HMAC_SECRET")
timestamp = raw_request.headers.get("x-pinless-timestamp")
signature = raw_request.headers.get("x-pinless-signature")
if not hmac_secret:
logger.debug("Skipping HMAC validation - PINLESS_HMAC_SECRET not set")
elif timestamp and signature:
message = timestamp + "." + raw_body_str
base64_decoded_secret = base64.b64decode(hmac_secret)
computed_signature = base64.b64encode(
hmac.new(base64_decoded_secret, message.encode(), "sha256").digest()
).decode()
if computed_signature != signature:
logger.error(f"Invalid signature. Expected {signature}, got {computed_signature}")
raise HTTPException(status_code=401, detail="Invalid signature")
else:
logger.debug("Skipping HMAC validation - no signature headers present")
if request.test == "test":
logger.debug("Test request received")
return {"status": "success", "message": "Test request received"}
dialin_settings = None
# these fields are camelCase in the request
required_fields = ["To", "From", "callId", "callDomain"]
if all(
field in request.dict() and request.dict()[field] is not None for field in required_fields
):
# transform from camelCase to snake_case because daily-python expects snake_case
dialin_settings = {
"From": request.From,
"To": request.To,
"call_id": request.callId,
"call_domain": request.callDomain,
# transform from camelCase to snake_case
}
logger.debug(f"Populated dialin_settings from request: {dialin_settings}")
daily_room_properties = {
"enable_dialout": request.dialout_settings is not None,
}
if dialin_settings is not None:
sip_config = {
"display_name": request.From,
"sip_mode": "dial-in",
"num_endpoints": 2 if request.call_transfer is not None else 1,
}
daily_room_properties["sip"] = sip_config
# Setting default expiry to 5 minutes from now
daily_room_properties["exp"] = int(time.time()) + (5 * 60)
logger.debug(f"Daily room properties: {daily_room_properties}")
payload = {
"createDailyRoom": True,
"dailyRoomProperties": daily_room_properties,
"body": {
"dialin_settings": dialin_settings,
"dialout_settings": request.dialout_settings,
"voicemail_detection": request.voicemail_detection,
"call_transfer": request.call_transfer,
},
}
pcc_api_key = os.getenv("PIPECAT_CLOUD_API_KEY")
agent_name = os.getenv("AGENT_NAME", "my-first-agent")
if not pcc_api_key:
raise HTTPException(status_code=500, detail="DAILY_API_KEY environment variable is not set")
headers = {"Authorization": f"Bearer {pcc_api_key}", "Content-Type": "application/json"}
url = f"https://api.pipecat.daily.co/v1/public/{agent_name}/start"
logger.debug(f"Making API call to Daily: {url} {headers} {payload}")
try:
response = requests.post(url, json=payload, headers=headers)
response.raise_for_status()
response_data = response.json()
logger.debug(f"Response: {response_data}")
return {
"status": "success",
"data": response_data,
"room_properties": daily_room_properties,
}
except requests.exceptions.HTTPError as e:
# Pass through the status code and error details from the Daily API
status_code = e.response.status_code
error_detail = e.response.json() if e.response.content else str(e)
logger.error(f"HTTP error: {error_detail}")
raise HTTPException(status_code=status_code, detail=error_detail)
except requests.exceptions.RequestException as e:
logger.error(f"Request error: {str(e)}")
raise HTTPException(status_code=500, detail=str(e))
if __name__ == "__main__":
try:
import uvicorn
uvicorn.run(app, host="0.0.0.0", port=7860)
except KeyboardInterrupt:
logger.info("Server stopped manually")

View File

@@ -0,0 +1,53 @@
# dependencies
/node_modules
/.pnp
.pnp.js
# testing
/coverage
# next.js
/.next/
/out/
# production
/build
# misc
.DS_Store
*.pem
# debug
npm-debug.log*
yarn-debug.log*
yarn-error.log*
.pnpm-debug.log*
# local env files
.env*.local
# vercel
.vercel
# typescript
*.tsbuildinfo
next-env.d.ts
# IDE specific files
.idea/
.vscode/
*.swp
*.swo
# Logs
logs
*.log
# OS generated files
.DS_Store
.DS_Store?
._*
.Spotlight-V100
.Trashes
ehthumbs.db
Thumbs.db

View File

@@ -0,0 +1,115 @@
# Next.js server for handling Daily PSTN/SIP Webhook
Next.js API routes for handling Daily PSTN/SIP Pipecat requests.
## Features
- API endpoint for handling Daily PSTN/SIP Pipecat requests
- HMAC signature validation
- Structured logging with Pino
- Support for dial-in and dial-out settings
- Voicemail detection and call transfer functionality
- Test request handling
## Setup
1. Clone the repository
2. Navigate to the `nextjs-webhook-server` directory:
```bash
cd nextjs-webhook-server
```
3. Install dependencies:
```bash
npm install
```
4. Create `.env.local` file with your credentials:
```bash
cp env.local.example .env.local
```
5. Update your `.env` with your secrets:
```bash
PIPECAT_CLOUD_API_KEY=pk_*
AGENT_NAME=my-first-agent
PINLESS_HMAC_SECRET=your_hmac_secret
LOG_LEVEL=info
```
### Running the server
Run the development server:
```bash
npm run dev
```
The server will run on `http://localhost:7860` and you can expose it via ngrok for testing:
```bash
`ngrok http 7860`
```
> Tip: Use a subdomain for a consistent URL (e.g. `ngrok http -subdomain=mydomain http://localhost:7860`)
## API Endpoints
### GET /api
Returns a simple "Hello, World!" message with a cute cat emoji to verify the server is running.
### POST /api/dial
Handles dial-in and dial-out requests for Pipecat Cloud.
#### Test Requests
The endpoint handles test requests when a webhook is configured. Send a request with `"Test": "test"` to verify your setup:
```json
{
"Test": "test"
}
```
#### Production Request Format
```json
{
// for dial-in from webhook
"To": "+14152251493",
"From": "+14158483432",
"callId": "string-contains-uuid",
"callDomain": "string-contains-uuid",
// for making a dial out to a phone or SIP
"dialout_settings": [
{ "phoneNumber": "+14158483432", "callerId": "purchased_phone_uuid" },
{ "sipUri": "sip:username@sip.hostname.com" }
]
}
```
## Deployment
The application is configured for Vercel deployment:
1. Push your code to a Git repository
2. Import your project in Vercel dashboard
3. Configure environment variables:
- `PIPECAT_CLOUD_API_KEY`
- `AGENT_NAME`
- `PINLESS_HMAC_SECRET`
- `LOG_LEVEL` (optional, defaults to 'info')
4. Deploy!
## Security
- HMAC signature validation for request authentication
- Environment variables for sensitive credentials
- Method validation (POST only for /dial)

View File

@@ -0,0 +1,4 @@
AGENT_NAME=my-first-agent
PIPECAT_CLOUD_API_KEY=your_daily_api_key
PINLESS_HMAC_SECRET=your_hmac_secret
LOG_LEVEL="info"

File diff suppressed because it is too large Load Diff

View File

@@ -0,0 +1,22 @@
{
"name": "my-daily-app",
"version": "0.1.0",
"private": true,
"scripts": {
"dev": "next dev -p 7860",
"build": "next build",
"start": "next start -p 7860",
"lint": "next lint"
},
"dependencies": {
"axios": "^1.6.0",
"next": "^14.0.0",
"pino": "^8.15.0",
"react": "^18.2.0",
"react-dom": "^18.2.0"
},
"devDependencies": {
"eslint": "^8.46.0",
"eslint-config-next": "^14.0.0"
}
}

View File

@@ -0,0 +1,175 @@
import { logger } from '../../lib/utils';
import axios from 'axios';
import crypto from 'crypto';
const validateSignature = (body, signature, timestamp, secret) => {
// Skip if any required fields are missing
if (!signature || !timestamp || !secret) {
logger.warn('Missing required fields for HMAC validation');
return true;
}
try {
const decodedSecret = Buffer.from(secret, 'base64');
const hmac = crypto.createHmac('sha256', decodedSecret);
const signatureData = `${timestamp}.${body}`;
const computedSignature = hmac.update(signatureData).digest('base64');
logger.debug('Signature validation:', {
timestamp,
signatureData: signatureData.substring(0, 50) + '...',
computedSignature,
receivedSignature: signature
});
return computedSignature === signature;
} catch (error) {
logger.error('Error validating signature:', error);
return true; // Allow request to proceed on error
}
};
export default async function handler(req, res) {
// Only allow POST requests
if (req.method !== 'POST') {
return res.status(405).json({ error: 'Method not allowed' });
}
try {
logger.info('Incoming request to /api/dial:');
logger.info(`Headers: ${JSON.stringify(req.headers)}`);
const rawBody = JSON.stringify(req.body);
logger.info(`Raw body: ${rawBody}`);
const signature = req.headers['x-pinless-signature'];
const timestamp = req.headers['x-pinless-timestamp'];
if (signature && timestamp) {
logger.info('Validating HMAC signature');
if (!validateSignature(rawBody, signature, timestamp, process.env.PINLESS_HMAC_SECRET)) {
logger.error('Invalid HMAC signature', { signature, timestamp });
return res.status(401).json({
error: 'Invalid signature',
message: 'Invalid HMAC signature'
});
}
} else {
logger.info('Skipping HMAC validation - no signature headers present');
}
// Extract request data
const {
Test: test,
To,
From,
callId,
callDomain,
dialout_settings,
voicemail_detection,
call_transfer
} = req.body;
// Handle test requests when a webhook is configured
if (test === 'test') {
logger.debug('Test request received');
return res.status(200).json({ status: 'success', message: 'Test request received' });
}
// Process dialin settings
let dialin_settings = null;
const requiredFields = ['To', 'From', 'callId', 'callDomain'];
if (requiredFields.every(field => req.body[field] !== undefined && req.body[field] !== null)) {
dialin_settings = {
// snake_case because pipecat expects this format
From,
To,
call_id: callId,
call_domain: callDomain,
};
logger.debug(`Populated dialin_settings from request: ${JSON.stringify(dialin_settings)}`);
}
// Set up Daily room properties
const daily_room_properties = {
enable_dialout: dialout_settings !== undefined && dialout_settings !== null,
exp: Math.floor(Date.now() / 1000) + (5 * 60), // 5 minutes from now
};
// Configure SIP if dialin settings are provided
if (dialin_settings !== null) {
const sip_config = {
display_name: From,
sip_mode: 'dial-in',
num_endpoints: call_transfer !== null ? 2 : 1,
};
daily_room_properties.sip = sip_config;
}
// Prepare payload for {service}/start API call
const payload = {
createDailyRoom: true,
dailyRoomProperties: daily_room_properties,
body: {
dialin_settings,
dialout_settings,
voicemail_detection,
call_transfer,
},
};
logger.debug(`Daily room properties: ${JSON.stringify(daily_room_properties)}`);
// Get Daily API key and agent name from environment variables
const pccApiKey = process.env.PIPECAT_CLOUD_API_KEY;
const agentName = process.env.AGENT_NAME || 'my-first-agent';
if (!pccApiKey) {
throw new Error('PIPECAT_CLOUD_API_KEY environment variable is not set');
}
// Set up headers for Daily API call
const headers = {
'Authorization': `Bearer ${pccApiKey}`,
'Content-Type': 'application/json',
};
const url = `https://api.pipecat.daily.co/v1/public/${agentName}/start`;
logger.debug(`Making API call to Daily: ${url} ${JSON.stringify(headers)} ${JSON.stringify(payload)}`);
try {
const response = await axios.post(url, payload, { headers });
logger.debug(`Response: ${JSON.stringify(response.data)}`);
return res.status(200).json({
status: 'success',
data: response.data,
room_properties: daily_room_properties,
});
} catch (error) {
if (error.response) {
// Pass through status code and error details from the Daily API
const statusCode = error.response.status;
const errorDetail = error.response.data || error.message;
logger.error(`HTTP error: ${JSON.stringify(errorDetail)}`);
return res.status(statusCode).json(errorDetail);
} else {
logger.error(`Request error: ${error.message}`);
return res.status(500).json({ error: error.message });
}
}
} catch (error) {
logger.error(`Unexpected error: ${error.message}`);
return res.status(500).json({ error: 'Internal server error', message: error.message });
}
}
// Configure body parser to preserve raw body text
export const config = {
api: {
bodyParser: {
sizeLimit: '1mb',
},
},
};

View File

@@ -0,0 +1,6 @@
import { logger } from '../../lib/utils';
export default function handler(req, res) {
logger.info('Received request to /api');
res.status(200).json({ message: 'Hello, World! from ᓚᘏᗢ' });
}

View File

@@ -0,0 +1,6 @@
module.exports = {
version: 2,
buildCommand: "next build",
outputDirectory: ".next",
cleanUrls: true
};

View File

@@ -159,8 +159,8 @@ class AzureTTSService(AzureBaseTTSService):
self._speech_config = SpeechConfig(
subscription=self._api_key,
region=self._region,
speech_recognition_language=self._settings["language"],
)
self._speech_config.speech_synthesis_language = self._settings["language"]
self._speech_config.set_speech_synthesis_output_format(
sample_rate_to_output_format(self.sample_rate)
)
@@ -254,8 +254,8 @@ class AzureHttpTTSService(AzureBaseTTSService):
self._speech_config = SpeechConfig(
subscription=self._api_key,
region=self._region,
speech_recognition_language=self._settings["language"],
)
self._speech_config.speech_synthesis_language = self._settings["language"]
self._speech_config.set_speech_synthesis_output_format(
sample_rate_to_output_format(self.sample_rate)
)

View File

@@ -158,7 +158,7 @@ class CartesiaTTSService(AudioContextWordTTSService):
voice_config["__experimental_controls"]["emotion"] = self._settings["emotion"]
msg = {
"transcript": text or " ", # Text must contain at least one character
"transcript": text,
"continue": continue_transcript,
"context_id": self._context_id,
"model_id": self.model_name,
@@ -287,7 +287,7 @@ class CartesiaTTSService(AudioContextWordTTSService):
self._context_id = str(uuid.uuid4())
await self.create_audio_context(self._context_id)
msg = self._build_msg(text=text or " ") # Text must contain at least one character
msg = self._build_msg(text=text)
try:
await self._get_websocket().send(msg)

View File

@@ -105,7 +105,7 @@ class OpenAITTSService(TTSService):
extra_body["instructions"] = self._instructions
async with self._client.audio.speech.with_streaming_response.create(
input=text or " ", # Text must contain at least one character
input=text,
model=self.model_name,
voice=VALID_VOICES[self._voice_id],
response_format="pcm",

View File

@@ -269,7 +269,8 @@ class TTSService(AIService):
filter.reset_interruption()
text = filter.filter(text)
await self.process_generator(self.run_tts(text))
if text:
await self.process_generator(self.run_tts(text))
await self.stop_processing_metrics()

View File

@@ -37,7 +37,7 @@ from pipecat.processors.frame_processor import FrameDirection, FrameProcessor
from pipecat.transports.base_transport import TransportParams
from pipecat.utils.time import nanoseconds_to_seconds
BOT_VAD_STOP_SECS = 0.3
BOT_VAD_STOP_SECS = 0.35
class BaseOutputTransport(FrameProcessor):