> ## Documentation Index
> Fetch the complete documentation index at: https://smallestai-ff1e543d.mintlify.site/llms.txt
> Use this file to discover all available pages before exploring further.
# Lightning v3.1 WebSocket
> The Lightning v3.1 WebSocket API provides real-time text-to-speech streaming capabilities with natural, expressive voice synthesis. This API uses WebSocket to deliver audio chunks as they're generated, enabling low-latency audio playback without waiting for the entire audio file to process.
## Overview
The Lightning v3.1 WebSocket API delivers state-of-the-art text-to-speech synthesis through a persistent WebSocket connection, providing ultra-low latency audio streaming for applications that demand real-time voice generation. Unlike traditional HTTP-based TTS APIs, this WebSocket implementation streams audio chunks as they're processed, significantly reducing perceived latency and enabling seamless user experiences.
Lightning v3.1 is a 44 kHz model that produces natural, expressive, and realistic speech, with support for voice cloning.
## Key Benefits
* **Ultra-Low Latency**: Audio chunks delivered as soon as they're generated
* **Real-Time Streaming**: Continuous audio delivery without waiting for complete text processing
* **Natural Speech**: Expressive synthesis that sounds realistic
* **Voice Cloning Support**: Compatible with cloned voices
* **Persistent Connection**: Maintains connection for multiple requests, reducing connection overhead
* **Interactive Applications**: Perfect for chatbots, voice assistants, and live communication systems
## Use Cases
* **Voice Assistants**: Real-time response generation for conversational AI
* **Interactive Chatbots**: Immediate audio feedback for user interactions
* **Live Streaming**: Real-time narration and commentary
* **Accessibility Tools**: Screen readers and text-to-speech applications
* **Gaming**: Dynamic voice generation for characters and narration
* **Customer Service**: Automated voice responses with natural speech patterns
## Concurrency and Rate Limits
This WebSocket API is subject to concurrency limits to ensure optimal performance for all users. Here's how it works:
* **1 Concurrency Unit** = 1 active TTS request that can be processed at any given time
* **5 WebSocket Connections** can be established per concurrency unit
* **Total Connections** = Your concurrency limit × 5
**Examples:**
* **1 concurrency** = Up to 5 WebSocket connections, but only 1 active request
* **3 concurrency** = Up to 15 WebSocket connections, but only 3 active requests simultaneously
* **5 concurrency** = Up to 25 WebSocket connections, but only 5 active requests simultaneously
While you can maintain multiple WebSocket connections, only your concurrency limit number of requests can be actively processed at once. Additional requests sent through any connection while at the concurrency limit will be rejected with an error.
For detailed information about concurrency limits, rate limiting, and best practices for handling these constraints, see our [Concurrency and Limits](concurrency-and-limits) documentation.
When multiple requests are sent simultaneously beyond your concurrency limit,
additional requests will be rejected with an error. Implement proper error
handling and request queuing to manage concurrency effectively.
## AsyncAPI
````yaml asyncapi-spec/lightning-v3.1-ws.json /waves/v1/lightning-v3.1/get_speech/stream
id: /waves/v1/lightning-v3.1/get_speech/stream
title: /waves/v1/lightning-v3.1/get_speech/stream
description: ''
servers:
- id: production
protocol: wss
host: api.smallest.ai
bindings: []
variables: []
address: /waves/v1/lightning-v3.1/get_speech/stream
parameters: []
bindings: []
operations:
- &ref_1
id: ttsRequest
title: Tts request
description: Send text-to-speech request
type: receive
messages:
- &ref_3
id: ttsRequest.message
contentType: application/json
payload:
- name: TTSRequest
type: object
properties:
- name: voice_id
type: string
description: The ID of the voice to use
required: true
- name: text
type: string
description: The text to convert to speech
required: true
- name: max_buffer_flush_ms
type: integer
description: >-
The maximum time (in ms) to wait for more input before
generating output. It flushes when either this time is reached
or enough input is received for optimal output—whichever comes
first. This is useful for input streams. Defaults to 0
required: false
- name: continue
type: boolean
description: >-
This setting controls whether the system should buffer and
wait for more input after receiving the current one. If not
set, it assumes no more input is coming.
required: false
- name: flush
type: boolean
description: >-
This setting controls whether the system should flush the
current buffer.
required: false
- name: complete_backoff_ms
type: number
description: >-
The time in ms to wait after the last chunk is sent before
sending the complete response. Default is 4000ms. Maximum is
10000ms.
required: false
- name: language
type: string
description: 'The language code, available options: `en`, `hi`'
enumValues:
- en
- hi
required: false
- name: sample_rate
type: integer
description: Audio sample rate in Hz
enumValues:
- 8000
- 16000
- 24000
- 44100
required: false
- name: speed
type: number
description: Speaking speed multiplier
required: false
headers: []
jsonPayloadSchema:
type: object
required:
- voice_id
- text
properties:
voice_id:
type: string
description: The ID of the voice to use
x-parser-schema-id:
text:
type: string
description: The text to convert to speech
x-parser-schema-id:
max_buffer_flush_ms:
type: integer
description: >-
The maximum time (in ms) to wait for more input before
generating output. It flushes when either this time is reached
or enough input is received for optimal output—whichever comes
first. This is useful for input streams. Defaults to 0
default: 0
maximum: 1000
minimum: 0
x-parser-schema-id:
continue:
type: boolean
description: >-
This setting controls whether the system should buffer and wait
for more input after receiving the current one. If not set, it
assumes no more input is coming.
default: false
x-parser-schema-id:
flush:
type: boolean
description: >-
This setting controls whether the system should flush the
current buffer.
default: false
x-parser-schema-id:
complete_backoff_ms:
type: number
description: >-
The time in ms to wait after the last chunk is sent before
sending the complete response. Default is 4000ms. Maximum is
10000ms.
default: 4000
minimum: 0
maximum: 10000
x-parser-schema-id:
language:
type: string
description: 'The language code, available options: `en`, `hi`'
default: en
enum:
- en
- hi
x-parser-schema-id:
sample_rate:
type: integer
description: Audio sample rate in Hz
default: 44100
enum:
- 8000
- 16000
- 24000
- 44100
x-parser-schema-id:
speed:
type: number
description: Speaking speed multiplier
minimum: 0.5
maximum: 2
default: 1
x-parser-schema-id:
examples:
- voice_id: ryan
text: Hello, this is a sample text to convert to speech.
max_buffer_flush_ms: 0
continue: false
flush: false
language: en
sample_rate: 24000
speed: 1
consistency: 0.5
enhancement: 1
similarity: 0
x-parser-schema-id:
title: TTSRequest
example: |-
{
"voice_id": "ryan",
"text": "Hello, this is a sample text to convert to speech.",
"max_buffer_flush_ms": 0,
"continue": false,
"flush": false,
"language": "en",
"sample_rate": 24000,
"speed": 1,
"consistency": 0.5,
"enhancement": 1,
"similarity": 0
}
bindings: []
extensions:
- id: x-parser-unique-object-id
value: ttsRequest.message
bindings: []
extensions: &ref_0
- id: x-parser-unique-object-id
value: /waves/v1/lightning-v3.1/get_speech/stream
- &ref_2
id: ttsResponse
title: Tts response
description: Receive audio stream chunks
type: send
messages:
- &ref_4
id: ttsResponse.message
contentType: application/json
payload:
- name: TTSResponse
type: object
properties:
- name: request_id
type: string
description: Unique identifier for the TTS request
required: false
- name: status
type: string
description: >-
Status of the TTS request, `chunk` indicates incoming audio
chunk, `complete` indicates completion.
enumValues:
- chunk
- complete
required: false
- name: data
type: object
required: false
properties:
- name: audio
type: string
description: Base64-encoded audio chunk
required: false
headers: []
jsonPayloadSchema:
type: object
properties:
request_id:
type: string
description: Unique identifier for the TTS request
x-parser-schema-id:
status:
type: string
enum:
- chunk
- complete
description: >-
Status of the TTS request, `chunk` indicates incoming audio
chunk, `complete` indicates completion.
x-parser-schema-id:
data:
type: object
properties:
audio:
type: string
description: Base64-encoded audio chunk
x-parser-schema-id:
x-parser-schema-id:
x-parser-schema-id:
title: TTSResponse
example: |-
{
"request_id": "",
"status": "",
"data": {
"audio": ""
}
}
bindings: []
extensions:
- id: x-parser-unique-object-id
value: ttsResponse.message
bindings: []
extensions: *ref_0
sendOperations:
- *ref_1
receiveOperations:
- *ref_2
sendMessages:
- *ref_3
receiveMessages:
- *ref_4
extensions:
- id: x-parser-unique-object-id
value: /waves/v1/lightning-v3.1/get_speech/stream
securitySchemes:
- id: bearerAuth
name: bearerAuth
type: http
description: Bearer token authentication
scheme: bearer
bearerFormat: JWT
extensions: []
````