Realtime voice and bounded audio

Updated 12 October 2026

Reviewed against Console v0.2.2. Available actions depend on project access and runtime configuration; check your installed CLI version for command differences.

Choose an enabled audio route#

The hosted gateway exposes speech synthesis, transcription and /api/v1/inference/realtime for supported realtime providers. Availability depends on the configured provider, model and workspace. Keep application bearer credentials on a trusted server or executor.

Provider and runtime choices#

The realtime client resolves its endpoint and model from configuration. Hosted provider-backed voice and a separately configured native runtime have different capabilities. Do not assume a voice model, audio rate or turn-control setting is interchangeable between them.

Manage one connection lifecycle#

Set up each connection once. Send audio in the format expected by that route, commit the input when using manual turn control, and request a response. Consume typed events until completion and handle cancellation or disconnection. A reconnect starts a fresh connection and needs its own setup.

Install the voice extra#

uv add 'stimulir[realtime]'

from stimulir.realtime import RealtimeClient, AudioDelta, ResponseDone

async with RealtimeClient(instructions="Answer briefly.").connect() as conn:
    await conn.setup()
    await conn.send_audio(pcm16_16k_mono)
    await conn.commit()
    await conn.create_response()
    async for event in conn.events():
        if isinstance(event, AudioDelta):
            consume_audio(event.pcm16)
        elif isinstance(event, ResponseDone):
            break

Provide authorized input audio and implement consume_audio for your playback or storage destination. The example shows the client’s PCM path; check the selected route’s format before connecting.

Bound work and inspect usage#

Use a bounded audio file or utterance while validating an integration. The CLI provides TTS, STT and realtime file-output commands below. Inspect trace and usage records after completion; audio cost and latency should be measured for the selected route.

Command catalogue#

Expand a command for its arguments. These commands follow the released CLI source; check stimulir --version and the command’s --help for your installation. Required flags are marked; some operations also validate combinations of inputs.