README
¶
SpeechKit Examples
These examples show the public framework surface that OSS consumers can import. They are intentionally component-level examples: copy the packages you need, not the entire desktop application.
library/: embeds the dictation recording and transcription pipeline with host-provided adapters.provider-catalog/: reads the current mode contracts and provider catalog used by host applications.embed-companion/: composes wake detections, explicit Hands-Free target routing, Assist requests, optional TTS, and runtime events through the public SDK surface.embed-tts/: demonstratespkg/speechkit/ttsProvider, Router, Service, and provider-kind routing.embed-event-bus/: publishes and consumes the additive wake, skill, companion, Voice-Agent, and TTS events.voice-agent/game-instructor/: end-to-end 15-minute Voice Agent embedded in a Go program (persona/role/sequence TOML + WebSocket client). Reference for the single-prompt "build a voice agent into my app" use case.voice-agent/provider-switching/: credential-free provider/profile/model selection using the publicpkg/speechkit/voiceagent/livedescriptors,ProviderIntent, and config helpers.
Agent prompts
Assist Voice Companion:
Add a SpeechKit Assist Voice Companion to this Go app. Use examples/embed-companion as the reference, import only pkg/speechkit/{companion,wakeword,assist,tts} plus pkg/speechkit for events, and wire companion.NewHandsFree with TargetMode: companion.TargetAssist. The host owns mic capture and playback. Do not import internal/*.
Voice Agent companion:
Build a hands-free SpeechKit Voice Agent companion into this app. Use companion.TargetVoiceAgent for wake activation and examples/voice-agent/game-instructor for the realtime session pattern. Use pkg/speechkit/client when talking to speechkit-server, or pkg/speechkit/agentkit when embedding a Go Voice Agent harness. Do not load the Windows client.
Dictation UI-assisted activation:
Add hands-free Dictation activation to this UI app with companion.TargetDictationUIAssisted. Wake can start/stop capture, but committing text must go through a visible target or explicit UI action. Do not synthesize TTS for Dictation.
Run an example from the repository root:
go run ./examples/provider-catalog
go run ./examples/embed-companion
go run ./examples/embed-tts
go run ./examples/embed-event-bus
go run ./examples/voice-agent/provider-switching -provider assemblyai
go run ./examples/voice-agent/provider-switching -provider auto -preferred openai,assemblyai -require-native-keyterms
Directories
¶
| Path | Synopsis |
|---|---|
|
assist
|
|
|
in-process
command
Example: fully in-process Assist — text (or speech) in, one useful result out, no SpeechKit server.
|
Example: fully in-process Assist — text (or speech) in, one useful result out, no SpeechKit server. |
|
Example: Using SpeechKit as a Go library for speech-to-text.
|
Example: Using SpeechKit as a Go library for speech-to-text. |
|
Example: reading SpeechKit's public mode and provider catalog.
|
Example: reading SpeechKit's public mode and provider catalog. |
|
voice-agent
|
|
|
game-instructor
command
Example: 15-minute Voice-Agent game instructor.
|
Example: 15-minute Voice-Agent game instructor. |
|
in-process
command
Example: fully in-process Voice Agent — no SpeechKit server in the path.
|
Example: fully in-process Voice Agent — no SpeechKit server in the path. |
|
provider-switching
command
Package main demonstrates provider/profile/model selection without live credentials.
|
Package main demonstrates provider/profile/model selection without live credentials. |
Click to show internal directories.
Click to hide internal directories.