类class agentscope.realtime

LocalAudioTransport

class LocalAudioTransport(TransportBase)

源码Source agentscope/realtime/_transport/_local.py:15–209

本地音频传输层，处理低延迟的实时语音与多模态交互。位于 agentscope.realtime 包，继承自 TransportBase。核心成员包括 start、close、incoming、send_audio、clear_audio、playout。

Audio callbacks run on PortAudio's thread, not the event loop: captured chunks are handed to the loop with call_soon_threadsafe and playback pulls from a lock-guarded buffer. Playout accounting lives in that callback, the same place a browser's AudioWorklet does it.

构造器摘要Constructor Summary

构造器Constructor	说明Description
`def __init__(input_sample_rate: int = 16000, output_sample_rate: int = 24000, input_device: int \| str \| None = None, ou…`	初始化 `LocalAudioTransport` 实例。 Initialize the transport.

方法摘要Method Summary

方法Method	说明Description
`async def start() -> None`async	返回启动。 Open both streams.
`async def close() -> None`async	关闭连接并释放相关资源。 Stop both streams and end :meth:`incoming`.
`async def incoming() -> AsyncIterator[TransportFrame]`async	返回 `incoming` 结果。 Yield captured chunks until :meth:`close`.
`async def send_audio(pcm: bytes, item_id: str) -> None`async	发送音频。 Queue assistant audio. A new item resets the playout counters.
`async def clear_audio() ->` `PlayoutPosition`async	清空音频。 Fade the head of the queue out, drop the rest, report the cut.
`def playout() ->` `PlayoutPosition`	返回 `playout` 结果。 The current playback position.

构造器详情Constructor Detail

LocalAudioTransport

def __init__(input_sample_rate: int = 16000, output_sample_rate: int = 24000, input_device: int | str | None = None, output_device: int | str | None = None, chunk_ms: int = 100, fade_ms: int = 30) -> None

初始化 LocalAudioTransport 实例。

Initialize the transport.

参数Parameters

名称Name	类型Type	说明Description
`input_sample_rate`默认default `16000`	int	输入 sample 速率（类型 `int`）。Capture rate; should match the model's input rate.
`output_sample_rate`默认default `24000`	int	输出 sample 速率（类型 `int`）。Playback rate; should match the model's output rate.
`input_device`默认default `None`	int \| str \| None	输入 device（类型 `int \| str \| None`）。`sounddevice` input device, default if `None`.
`output_device`默认default `None`	int \| str \| None	输出 device（类型 `int \| str \| None`）。`sounddevice` output device, default if `None`.
`chunk_ms`默认default `100`	int	分块 ms（类型 `int`）。Capture chunk length handed to :meth:`incoming`.
`fade_ms`默认default `30`	int	fade ms（类型 `int`）。Fade-out applied by :meth:`clear_audio` to avoid a click.

源码Source agentscope/realtime/_transport/_local.py:24–68

方法详情Method Detail

start

async def start() -> None

async

返回启动。

Open both streams.

返回Returns None

无返回值。No value is returned.

源码Source agentscope/realtime/_transport/_local.py:74–98

close

async def close() -> None

async

关闭连接并释放相关资源。

Stop both streams and end :meth:incoming.

返回Returns None

无返回值。No value is returned.

源码Source agentscope/realtime/_transport/_local.py:100–114

incoming

async def incoming() -> AsyncIterator[TransportFrame]

async

返回 incoming 结果。

Yield captured chunks until :meth:close.

返回Returns AsyncIterator[TransportFrame]

以异步流的形式返回TransportFrame。Returns a stream of transport frame.

源码Source agentscope/realtime/_transport/_local.py:120–126

send_audio

async def send_audio(pcm: bytes, item_id: str) -> None

async

发送音频。

Queue assistant audio.

参数Parameters

名称Name	类型Type	说明Description
`pcm`	bytes	pcm（类型 `bytes`）。The pcm.
`item_id`	str	条目标识（类型 `str`）。The item id.

返回Returns None

无返回值。No value is returned.

源码Source agentscope/realtime/_transport/_local.py:154–162

clear_audio

async def clear_audio() -> PlayoutPosition

async

清空音频。

Fade the head of the queue out, drop the rest, report the cut.

返回Returns PlayoutPosition

返回PlayoutPosition。Returns playout position.

源码Source agentscope/realtime/_transport/_local.py:186–196

playout

def playout() -> PlayoutPosition

返回 playout 结果。

The current playback position.

返回Returns PlayoutPosition

返回PlayoutPosition。Returns playout position.

源码Source agentscope/realtime/_transport/_local.py:198–201

继承成员Inherited Members

继承自Inherited from TransportBase

__aenter__, __aexit__, input_sample_rate, output_sample_rate