voice-goio docs

Integrating

Adding it

go get github.com/mrlm-net/[email protected]

Requires Go 1.27 or newer. The go directive in go.mod is a floor on the consumer’s toolchain, not a target, so it is set to the version the library is actually developed and tested on. A machine on an older toolchain will fetch 1.27 automatically on the first build.

Developing alongside an application

While the library and the application are changing together, point the app at the working copy rather than at a tag — a go.work beside both checkouts is the least intrusive way, because it leaves the app’s go.mod alone:

go 1.27

use (
    ./mycrew-online/app // or ./your-app
    ./voice-goio
)

Remove it, or drop the replace, before a release build.

What to wire first

Playback. tts.Open → pool.Assign → norm.Spoken → chain.Apply → player.Play, and drain player.Events() (see The pipeline), or let the speaker do it. This works on macOS with no downloads, so the integration can be proven before any Windows work.
Settings. player.Devices() for the output device; on Windows, sapi.InputDevices() for the microphone and sapi.Engines() to show whether a speech pack is installed at all.
Voices. Piper and a voice pack from the app’s first-run screen.
Recognition. stt/fake on the development machine, stt/sapi on Windows. Same interface, so the wiring is written once. See Recognition.

Versions

Tag	Contents
`v0.1.x`	Everything platform-independent, complete and tested. Windows backends compile and are wired, but have not run against hardware.
`v0.2.0`	Recognition failure reasons (`TagReason`), session recording, emergency and deviation phraseology.
`v0.3.0`	Departure clearance readbacks, WAV input for the Windows recogniser so the corpus runs unattended, and a fix for models downloading into the working directory.
`v0.4.0`	Business Source License 1.1 (non-commercial; Apache-2.0 four years after each release). Taxiway letters after “via” are spelled.
`v0.5.0`	`speaker`: the radio as applications speak it (voices per position and crew, one frequency, the queue, gaps, the ATIS broadcast). Female share in the voice pool; piper sentinel cut at its silence.
`v0.6.0`	The intercom, explicit voices (`Utterance.Voice`), `voices.Dir`/`Installed`, application grammars (`grammar.Commands`).
`v0.7.0`	The Czech controller voice through RP phonemes; `speaker.Options.Exclude`.
`v0.7.1`	Every accent model reads English through RP phonemes.
`v0.8.0`	Voice packs (`en`, `all`) and `Manifest.WritePack`.
`v0.9.0`	Controller accents opt-in (`speaker.Options.Accents`, off by default).
`v0.10.0`	`PackCore`, an installer’s default pack.
`v0.11.0`	`voices.InstallPack`, `Manifest.PackSize`, `piper.Install`.
`v0.11.1`	Fix: piper says a text of several sentences whole.
`v0.11.2`	Cabin chimes on the intercom.
`v0.12.0`	An output device per channel; the cabin PA channel and `PAChain`.
`v0.12.1`	Fix: one voice says one line at a time across channels.
`v0.13.0`	One controller, one voice on every frequency it works (`Utterance.Controller`).
`v0.13.1`	`Pool.AssignCrew`: crews never speak in a controller’s voice.

The full notes are in the changelog.

License

Business Source License 1.1 for versions after v0.3.1. Non-commercial use is free: personal and hobby use, the flight-simulation community, education, research and non-profits. Commercial use, such as a paid add-on or product, a paid service or use inside a business, needs a separate licence: write to [email protected]. Each version becomes Apache-2.0 four years after it is published, and versions up to and including v0.3.1 remain under Apache-2.0.