类class agentscope.realtime

Parameters

class Parameters(RealtimeModelBase.Parameters)

源码Source agentscope/realtime/_gemini/_model.py:43–66

参数，处理低延迟的实时语音与多模态交互。位于 agentscope.realtime 包，继承自 RealtimeModelBase.Parameters。

Tuneables surfaced to the UI. Declared in agentscope.realtime. Extends RealtimeModelBase.Parameters.

属性摘要Attribute Summary

属性Attribute	类型Type	说明Description	值Value
`voice`	str	语音字段。	`Field(default='Puck', title='Voice')`
`turn_detection`	Literal['automatic', 'none']	`turn_detection` 字段。	Field(default='automatic', title='Turn Detection', description='``none`` hands endpointing to the caller.')
`start_sensitivity`	Literal['HIGH', 'LOW']	`start_sensitivity` 字段。	`Field(default='HIGH', description='How eagerly speech is detected as started.')`
`end_sensitivity`	Literal['HIGH', 'LOW']	`end_sensitivity` 字段。	`Field(default='HIGH', description='How eagerly speech is detected as ended.')`
`prefix_padding_ms`	int	`prefix_padding_ms` 字段。	`Field(default=300, ge=0)`
`silence_duration_ms`	int	`silence_duration_ms` 字段。	`Field(default=800, ge=0)`
`input_audio_transcription`	bool	`input_audio_transcription` 字段。	`Field(default=True, title='Transcribe Input', description="Whether to transcribe the user's speech.")`

str = Field(default='Puck', title='Voice')

语音字段。声明为 str。

Voice

源码Source agentscope/realtime/_gemini/_model.py:46

Literal['automatic', 'none'] = Field(default='automatic', title='Turn Detection', description='``none`` hands endpointing to the caller.')

turn_detection 字段。声明为 Literal['automatic', 'none']。

none hands endpointing to the caller.

源码Source agentscope/realtime/_gemini/_model.py:47

Literal['HIGH', 'LOW'] = Field(default='HIGH', description='How eagerly speech is detected as started.')

start_sensitivity 字段。声明为 Literal['HIGH', 'LOW']。

How eagerly speech is detected as started.

源码Source agentscope/realtime/_gemini/_model.py:52

Literal['HIGH', 'LOW'] = Field(default='HIGH', description='How eagerly speech is detected as ended.')

end_sensitivity 字段。声明为 Literal['HIGH', 'LOW']。

How eagerly speech is detected as ended.

源码Source agentscope/realtime/_gemini/_model.py:56

int = Field(default=300, ge=0)

prefix_padding_ms 字段。声明为 int。

The prefix padding ms field.

源码Source agentscope/realtime/_gemini/_model.py:60

int = Field(default=800, ge=0)

silence_duration_ms 字段。声明为 int。

The silence duration ms field.

源码Source agentscope/realtime/_gemini/_model.py:61

bool = Field(default=True, title='Transcribe Input', description="Whether to transcribe the user's speech.")

input_audio_transcription 字段。声明为 bool。

Whether to transcribe the user's speech.

源码Source agentscope/realtime/_gemini/_model.py:62