Custom

Bring your own TTS engine by implementing TRTC's custom streaming protocol. This option gives you full control over the synthesis pipeline — use your proprietary model, on-premise deployment, or any third-party service not natively supported. Choose this if you have specialized voice requirements or need to integrate an in-house TTS solution.
Usage
To use a custom TTS engine, pass the following JSON in the TTSConfig field of the StartAIConversation API. Your TTS service must implement the TRTC custom TTS streaming protocol:
// json — TTSConfig
{
  "TTSType": "custom",
  "APIKey": "<your_api_key>",
  "APIUrl": "http://0.0.0.0:8080/stream-audio",
  "AudioFormat": "wav",
  "SampleRate": 16000,
  "AudioChannel": 1
}
For the complete TTSConfig parameter reference, see the Text-to-Speech Configuration.
Parameter reference
Field
Type
Required
Description
TTSType
String
Yes
Fixed value: "custom".
APIKey
String
Yes
API key for authentication with your TTS service.
APIUrl
String
Yes
Your TTS service endpoint URL.
AudioFormat
String
No
Output audio format. Currently supports: pcm, wav. Default: wav.
SampleRate
Integer
No
Audio sample rate. Default: 16000 (16 kHz). Recommended: 16000.
AudioChannel
Integer
No
Number of audio channels. 1 (mono) or 2 (stereo). Default: 1.
Custom TTS streaming protocol
Your TTS service must implement the following HTTP streaming protocol.
Calling Method
POST: http://xxxxxxxxxxxx/api/v1/tts/stream
Example
HTTP Request Header
Content-Type: application/json;charset=UTF-8
X-Task-Id: task_id_value
X-Rquest-Id: request_id
X-Sdk-App-Id: SdkAppId
X-User-Id: UserId
X-Room-Id: RoomId 
X-Room-Id-Type: "0" 
Authorization: Bearer "API-KEY" 
Request
{
  "Text": "Hello, world! This is a test for the streaming TTS API.",
  "Format": "wav",
  "SampleRate": 16000,
  "Channel": 1
}
HTTP Request Header
Field
Description
Content-Type
application/json
charset
UTF-8
X-Task-Id
ID of a conversation task.
X-Rquest-Id
ID of a request. Retries will carry the same RequestID.
X-Sdk-App-Id
AppId for the SDK.
X-User-Id
User ID.
X-Room-Id
Room ID.
X-Room-Id-Type
Room ID type. 0: numeric room number, 1: string room number.
Authorization
Authentication, in the format of Bearer "API-KEY".
Request
Parameter
Required
Type
Description
Text
Yes
String
Speech text.
Format
No
String
The desired audio format for output, such as mp3, ogg_opus, pcm, and wav. The default value is wav. Only pcm and wav are currently supported.
SampleRate
No
Integer
Audio sampling rate. The default value is 16000 (16k), with a recommended value of 16000.
Channel
No
Integer
Audio channel. Valid values: 1, 2. The default value is 1.
Response
The value of Content-Type needs to be used to determine whether the TTS is successful.
If successful, binary speech is returned. The Content-Type for different audio formats is as follows, and Transfer-Encoding: chunked should be set in the HTTP response header.
Audio Format
Content-Type
mp3
audio/mpeg
ogg_opus
audio/ogg
pcm
audio/L16
wav
audio/wav
If failed, the JSON result is returned with the header information: Content-type: application/json. The response is:
  {
  "error": {
    "code": "ERROR_CODE",
    "message": "A description of the error"
  }
}
Next step: StartAIConversation API Reference﻿

Field	Type	Required	Description
`TTSType`	String	Yes	Fixed value: `"custom"`.
`APIKey`	String	Yes	API key for authentication with your TTS service.
`APIUrl`	String	Yes	Your TTS service endpoint URL.
`AudioFormat`	String	No	Output audio format. Currently supports: `pcm`, `wav`. Default: `wav`.
`SampleRate`	Integer	No	Audio sample rate. Default: 16000 (16 kHz). Recommended: 16000.
`AudioChannel`	Integer	No	Number of audio channels. `1` (mono) or `2` (stereo). Default: 1.

Field	Description
Content-Type	application/json
charset	UTF-8
X-Task-Id	ID of a conversation task.
X-Rquest-Id	ID of a request. Retries will carry the same RequestID.
X-Sdk-App-Id	AppId for the SDK.
X-User-Id	User ID.
X-Room-Id	Room ID.
X-Room-Id-Type	Room ID type. 0: numeric room number, 1: string room number.
Authorization	Authentication, in the format of Bearer "API-KEY".

Parameter	Required	Type	Description
Text	Yes	String	Speech text.
Format	No	String	The desired audio format for output, such as mp3, ogg_opus, pcm, and wav. The default value is wav. Only pcm and wav are currently supported.
SampleRate	No	Integer	Audio sampling rate. The default value is 16000 (16k), with a recommended value of 16000.
Channel	No	Integer	Audio channel. Valid values: 1, 2. The default value is 1.

Audio Format	Content-Type
mp3	audio/mpeg
ogg_opus	audio/ogg
pcm	audio/L16
wav	audio/wav