xiaozhi

package

v1.4.0 Latest Latest Go to latest Published: Jun 8, 2026 License: MIT Imports: 16 Imported by: 0

Details

Valid go.mod file
Redistributable license
Tagged version
Stable version
Learn more about best practices

Repository

github.com/LingByte/lingllm

Links

Open Source Insights

Documentation ¶

Overview ¶

Package xiaozhi implements the xiaozhi-esp32 WebSocket voice protocol.

Device / Browser ── xiaozhi WS ──► Server ── dialog WS ──► Dialog (LLM)

Index ¶

Constants
func MakeAbortConfirm(sessionID string) []byte
func MakeError(message string, fatal bool) []byte
func MakeLLMReply(text string) []byte
func MakePongReply(sessionID string) []byte
func MakeSTTReply(sessionID, text string) []byte
func MakeTTSStateReply(sessionID, state, codec string) []byte
func MakeTTSStateReplyFrames(sessionID, state, codec string, frameMs int) []byte
func MakeWelcomeReply(sessionID string, ap AudioParams) []byte
func MergeHelloAudio(h *AudioParams)
func ParseTextFrame(raw []byte) (string, error)
type AudioParams
- func DefaultHelloAudio() AudioParams
type HelloMessage
type ListenMessage
type MessageHandler
type RealtimeAgentFactory
type Server
- func NewServer(cfg ServerConfig) (*Server, error)
- func (s *Server) Handle(w http.ResponseWriter, r *http.Request)
type ServerConfig

Constants ¶

View Source

const (
	MsgHello         = "hello"
	MsgListen        = "listen"
	MsgAbort         = "abort"
	MsgPing          = "ping"
	RespHello        = "hello"
	RespPong         = "pong"
	RespSTT          = "stt"
	RespTTS          = "tts"
	RespError        = "error"
	RespAbortConfirm = "abort"
)

View Source

const (
	ListenStart  = "start"
	ListenStop   = "stop"
	ListenDetect = "detect"
)

View Source

const (
	AudioFormatOpus = "opus"
	AudioFormatPCM  = "pcm"
)

View Source

const (
	// ModePipeline: ASR + dialog-plane WebSocket (LLM) + TTS.
	ModePipeline = "pipeline"
	// ModeRealtime: pkg/realtime multimodal agent (e.g. Qwen-Omni).
	ModeRealtime = "realtime"
)

Variables ¶

This section is empty.

Functions ¶

func MakeAbortConfirm ¶

func MakeAbortConfirm(sessionID string) []byte

func MakeError ¶

func MakeError(message string, fatal bool) []byte

func MakeLLMReply ¶

func MakeLLMReply(text string) []byte

func MakePongReply ¶

func MakePongReply(sessionID string) []byte

func MakeSTTReply ¶

func MakeSTTReply(sessionID, text string) []byte

func MakeTTSStateReply ¶

func MakeTTSStateReply(sessionID, state, codec string) []byte

func MakeTTSStateReplyFrames ¶

func MakeTTSStateReplyFrames(sessionID, state, codec string, frameMs int) []byte

func MakeWelcomeReply ¶

func MakeWelcomeReply(sessionID string, ap AudioParams) []byte

func MergeHelloAudio ¶

func MergeHelloAudio(h *AudioParams)

func ParseTextFrame ¶

func ParseTextFrame(raw []byte) (string, error)

Types ¶

type AudioParams ¶

type AudioParams struct {
	Format        string `json:"format,omitempty"`
	Codec         string `json:"codec,omitempty"`
	SampleRate    int    `json:"sample_rate,omitempty"`
	Channels      int    `json:"channels,omitempty"`
	FrameDuration int    `json:"frame_duration,omitempty"`
	BitDepth      int    `json:"bit_depth,omitempty"`
}

func DefaultHelloAudio ¶

func DefaultHelloAudio() AudioParams

type HelloMessage ¶

type HelloMessage struct {
	Type        string                 `json:"type"`
	Version     int                    `json:"version,omitempty"`
	Transport   string                 `json:"transport,omitempty"`
	Features    map[string]interface{} `json:"features,omitempty"`
	AudioParams *AudioParams           `json:"audio_params,omitempty"`
	Mode        string                 `json:"mode,omitempty"`
}

type ListenMessage ¶

type ListenMessage struct {
	Type  string `json:"type"`
	State string `json:"state"`
	Mode  string `json:"mode,omitempty"`
}

type MessageHandler ¶ added in v1.3.0

type MessageHandler func(ctx context.Context, session *wsSession, raw []byte) error

MessageHandler is a custom message handler for extensibility.

type RealtimeAgentFactory ¶

type RealtimeAgentFactory interface {
	NewAgent(ctx context.Context, callID string, onEvent func(realtime.Event)) (realtime.Agent, int, int, error)
}

RealtimeAgentFactory builds a started realtime.Agent for one xiaozhi session.

type Server ¶

type Server struct {
	// contains filtered or unexported fields
}

Server accepts xiaozhi WebSocket connections.

func NewServer ¶

func NewServer(cfg ServerConfig) (*Server, error)

NewServer validates cfg.

func (*Server) Handle ¶

func (s *Server) Handle(w http.ResponseWriter, r *http.Request)

Handle upgrades the HTTP connection to the xiaozhi protocol.

type ServerConfig ¶

type ServerConfig struct {
	SessionFactory       transport.SessionFactory
	RealtimeFactory      RealtimeAgentFactory
	DialogWSURL          string
	CallIDPrefix         string
	ResolveDevicePayload func(ctx context.Context, deviceID string) ([]byte, error)
	ConfigureClient      func(*gateway.ClientConfig)
	OnSessionStart       func(ctx context.Context, callID, deviceID string)
	OnSessionEnd         func(ctx context.Context, callID, reason string)
	// ConfigureSession allows third-party extensions to register custom message handlers
	ConfigureSession func(session *wsSession)
}

ServerConfig wires the xiaozhi adapter.

Source Files ¶

View all Source files

?	: This menu
/	: Search site
f or F	: Jump to
y or Y	: Canonical URL