你当前正在访问 Microsoft Azure Global Edition 技术文档网站。 如果需要访问由世纪互联运营的 Microsoft Azure 中国技术文档网站,请访问 https://docs.azure.cn

SpeechRecognizer class

从麦克风、文件或其他音频输入流执行语音识别,并获取转录文本作为结果。

Extends

构造函数

SpeechRecognizer(SpeechConfig, AudioConfig)

SpeechRecognizer 构造函数。

属性

authorizationToken

获取用于与服务通信的授权令牌。

canceled

事件已取消,表明识别过程中发生了错误。

endpointId

获取用于语音识别的自定义语音模型的终结点 ID。

internalData
outputFormat

获取识别的输出格式。

properties

为此 SpeechRecognizer 定义的属性及其值的集合。

recognized

已识别的事件表示已收到最终识别结果。

recognizing

识别事件表示收到中间识别结果的信号。

speechRecognitionLanguage

获取识别的口语。

telemetryEnabled

此方法返回遥测设置的当前状态。

继承属性

sessionStarted

定义会话启动事件的事件处理程序。

sessionStopped

定义会话停止事件的事件处理程序。

speechEndDetected

定义语音停止事件的事件处理程序。

speechStartDetected

定义语音启动事件的事件处理程序。

方法

close(() => void, (error: string) => void)

关闭此类的实例持有的所有外部资源。

dispose(boolean)

释放 对象持有的任何资源。

FromConfig(SpeechConfig, AutoDetectSourceLanguageConfig, AudioConfig)

SpeechRecognizer 构造函数。

recognizeOnceAsync((e: SpeechRecognitionResult) => void, (e: string) => void)

启动语音识别,并在识别第一个言语后停止。 该任务返回作为结果的识别文本。 注意:识别第一个言语时,RecognizeOnceAsync () 返回,因此它仅适用于单次识别(如命令或查询)。 对于长时间运行的识别,请改用 StartContinuousRecognitionAsync () 。

startContinuousRecognitionAsync(() => void, (e: string) => void)

启动语音识别,直到调用 stopContinuousRecognitionAsync () 。 用户必须订阅事件才能接收识别结果。

startKeywordRecognitionAsync(KeywordRecognitionModel, () => void, (e: string) => void)

使用关键字 (keyword) 发现启动语音识别,直到调用 stopKeywordRecognitionAsync () 。 用户必须订阅事件才能接收识别结果。 注意:关键字查找功能仅在语音设备 SDK 上可用。 此功能目前不包括在 SDK 本身中。

stopContinuousRecognitionAsync(() => void, (e: string) => void)

停止连续语音识别。

stopKeywordRecognitionAsync(() => void)

停止连续语音识别。 注意:关键字查找功能仅在语音设备 SDK 上可用。 此功能目前不包括在 SDK 本身中。

继承的方法

enableTelemetry(boolean)

此方法全局启用或禁用遥测。

构造函数详细信息

SpeechRecognizer(SpeechConfig, AudioConfig)

SpeechRecognizer 构造函数。

new SpeechRecognizer(speechConfig: SpeechConfig, audioConfig?: AudioConfig)

参数

speechConfig
SpeechConfig

此识别器的初始属性集

audioConfig
AudioConfig

与识别器关联的可选音频配置

属性详细信息

authorizationToken

获取用于与服务通信的授权令牌。

string authorizationToken

属性值

string

授权令牌。

canceled

事件已取消,表明识别过程中发生了错误。

public canceled: (sender: Recognizer, event: SpeechRecognitionCanceledEventArgs) => void

属性值

(sender: Recognizer, event: SpeechRecognitionCanceledEventArgs) => void

endpointId

获取用于语音识别的自定义语音模型的终结点 ID。

string endpointId

属性值

string

用于语音识别的自定义语音模型的终结点 ID。

internalData

object internalData

属性值

object

outputFormat

获取识别的输出格式。

OutputFormat outputFormat

属性值

识别的输出格式。

properties

为此 SpeechRecognizer 定义的属性及其值的集合。

PropertyCollection properties

属性值

为此 SpeechRecognizer 定义的属性及其值的集合。

recognized

已识别的事件表示已收到最终识别结果。

public recognized: (sender: Recognizer, event: SpeechRecognitionEventArgs) => void

属性值

(sender: Recognizer, event: SpeechRecognitionEventArgs) => void

recognizing

识别事件表示收到中间识别结果的信号。

public recognizing: (sender: Recognizer, event: SpeechRecognitionEventArgs) => void

属性值

(sender: Recognizer, event: SpeechRecognitionEventArgs) => void

speechRecognitionLanguage

获取识别的口语。

string speechRecognitionLanguage

属性值

string

识别的口语。

telemetryEnabled

此方法返回遥测设置的当前状态。

static boolean telemetryEnabled

属性值

boolean

如果启用遥测,则为 true;否则为 false。

继承属性详细信息

sessionStarted

定义会话启动事件的事件处理程序。

public sessionStarted: (sender: Recognizer, event: SessionEventArgs) => void

属性值

(sender: Recognizer, event: SessionEventArgs) => void

继承自Recognizer.sessionStarted

sessionStopped

定义会话停止事件的事件处理程序。

public sessionStopped: (sender: Recognizer, event: SessionEventArgs) => void

属性值

(sender: Recognizer, event: SessionEventArgs) => void

继承自Recognizer.sessionStopped

speechEndDetected

定义语音停止事件的事件处理程序。

public speechEndDetected: (sender: Recognizer, event: RecognitionEventArgs) => void

属性值

(sender: Recognizer, event: RecognitionEventArgs) => void

继承自Recognizer.speechEndDetected

speechStartDetected

定义语音启动事件的事件处理程序。

public speechStartDetected: (sender: Recognizer, event: RecognitionEventArgs) => void

属性值

(sender: Recognizer, event: RecognitionEventArgs) => void

继承自Recognizer.speechStartDetected

方法详细信息

close(() => void, (error: string) => void)

关闭此类的实例持有的所有外部资源。

function close(cb?: () => void, errorCb?: (error: string) => void)

参数

cb

() => void

errorCb

(error: string) => void

dispose(boolean)

释放 对象持有的任何资源。

function dispose(disposing: boolean): Promise<void>

参数

disposing

boolean

如果释放 对象,则为 true。

返回

Promise<void>

FromConfig(SpeechConfig, AutoDetectSourceLanguageConfig, AudioConfig)

SpeechRecognizer 构造函数。

static function FromConfig(speechConfig: SpeechConfig, autoDetectSourceLanguageConfig: AutoDetectSourceLanguageConfig, audioConfig?: AudioConfig): SpeechRecognizer

参数

speechConfig
SpeechConfig

此识别器的初始属性集

autoDetectSourceLanguageConfig
AutoDetectSourceLanguageConfig

与识别器关联的源语言检测配置

audioConfig
AudioConfig

与识别器关联的可选音频配置

返回

recognizeOnceAsync((e: SpeechRecognitionResult) => void, (e: string) => void)

启动语音识别,并在识别第一个言语后停止。 该任务返回作为结果的识别文本。 注意:识别第一个言语时,RecognizeOnceAsync () 返回,因此它仅适用于单次识别(如命令或查询)。 对于长时间运行的识别,请改用 StartContinuousRecognitionAsync () 。

function recognizeOnceAsync(cb?: (e: SpeechRecognitionResult) => void, err?: (e: string) => void)

参数

cb

(e: SpeechRecognitionResult) => void

接收 SpeechRecognitionResult 的回调。

err

(e: string) => void

发生错误时调用的回调。

startContinuousRecognitionAsync(() => void, (e: string) => void)

启动语音识别,直到调用 stopContinuousRecognitionAsync () 。 用户必须订阅事件才能接收识别结果。

function startContinuousRecognitionAsync(cb?: () => void, err?: (e: string) => void)

参数

cb

() => void

识别开始后调用回调。

err

(e: string) => void

发生错误时调用的回调。

startKeywordRecognitionAsync(KeywordRecognitionModel, () => void, (e: string) => void)

使用关键字 (keyword) 发现启动语音识别,直到调用 stopKeywordRecognitionAsync () 。 用户必须订阅事件才能接收识别结果。 注意:关键字查找功能仅在语音设备 SDK 上可用。 此功能目前不包括在 SDK 本身中。

function startKeywordRecognitionAsync(model: KeywordRecognitionModel, cb?: () => void, err?: (e: string) => void)

参数

model
KeywordRecognitionModel

指定要识别关键字 (keyword) 的关键字 (keyword) 识别模型。

cb

() => void

识别开始后调用的回调。

err

(e: string) => void

发生错误时调用的回调。

stopContinuousRecognitionAsync(() => void, (e: string) => void)

停止连续语音识别。

function stopContinuousRecognitionAsync(cb?: () => void, err?: (e: string) => void)

参数

cb

() => void

在识别停止后调用的回调。

err

(e: string) => void

发生错误时调用的回调。

stopKeywordRecognitionAsync(() => void)

停止连续语音识别。 注意:关键字查找功能仅在语音设备 SDK 上可用。 此功能目前不包括在 SDK 本身中。

function stopKeywordRecognitionAsync(cb?: () => void)

参数

cb

() => void

在识别停止后调用的回调。

继承的方法详细信息

enableTelemetry(boolean)

此方法全局启用或禁用遥测。

static function enableTelemetry(enabled: boolean)

参数

enabled

boolean

遥测收集的全局设置。 如果设置为 true,则会收集麦克风错误、识别错误等遥测信息并将其发送给 Microsoft。 如果设置为 false,则不会向 Microsoft 发送遥测数据。

继承自Recognizer.enableTelemetry