/// <summary>
/// 创建实时音频流识别对象.
/// </summary>
/// <returns>成功返回SpeechTranscriberRequest对象, 否则返回NULL.</returns>
public SpeechTranscriberRequest CreateTranscriberRequest()
{
IntPtr request = NativeMethods.NlsCreateTranscriberRequest();
SpeechTranscriberRequest STrequest = new SpeechTranscriberRequest();
STrequest.native_request = request;
return(STrequest);
}
/// <summary>
/// 设置一句话结束回调函数, 检测到了一句话的结束时, sdk内部线程上报该回调.
/// </summary>
/// <param name="request">
/// CreateTranscriberRequest所建立的request对象.
/// </param>
/// <param name="callback">
/// 用户传入的回调函数.
/// </param>
/// <param name="para">
/// 用户对象.
/// </param>
/// <returns></returns>
public void SetOnSentenceEnd(
SpeechTranscriberRequest request, CallbackDelegate callback, object para = null)
{
SpeechParamStruct user_param = new SpeechParamStruct();
user_param.user = para;
user_param.callback = callback;
user_param.nlsEvent = new NLS_EVENT_STRUCT();
IntPtr toCppParam = Marshal.AllocHGlobal(Marshal.SizeOf(user_param));
Marshal.StructureToPtr(user_param, toCppParam, false);
NativeMethods.STOnSentenceEnd(request.native_request, onSentenceEnd, (IntPtr)toCppParam);
return;
}
/// <summary>
/// 会与服务端确认关闭, 正常停止实时音频流识别操作. 异步操作, 失败返回TaskFailed.
/// </summary>
/// <param name="request">
/// CreateTranscriberRequest所建立的request对象.
/// </param>
/// <returns>成功则返回0, 否则返回-1.</returns>
public int Stop(SpeechTranscriberRequest request)
{
return(NativeMethods.STstop(request.native_request));
}
/// <summary>
/// 设置用户自定义ws阶段http header参数
/// </summary>
/// <param name="request">
/// CreateTranscriberRequest所建立的request对象.
/// </param>
/// <param name="value">
///
/// </param>
/// <returns>成功则返回0, 否则返回-1.</returns>
public int AppendHttpHeaderParam(SpeechTranscriberRequest request, string key, string value)
{
return(NativeMethods.STappendHttpHeaderParam(request.native_request, key, value));
}
/// <summary>
/// 设置用户自定义参数
/// </summary>
/// <param name="request">
/// CreateTranscriberRequest所建立的request对象.
/// </param>
/// <param name="value">
/// value json格式字符串, 类似"{\"network\":{\"ip\":\"100.101.102.103\"}}".
/// </param>
/// <returns>成功则返回0, 否则返回-1.</returns>
public int SetContextParam(SpeechTranscriberRequest request, string value)
{
return(NativeMethods.STsetContextParam(request.native_request, value));
}
/// <summary>
/// 设置输出文本的编码格式
/// </summary>
/// <param name="request">
/// CreateTranscriberRequest所建立的request对象.
/// </param>
/// <param name="value">
/// value 编码格式 UTF-8 or GBK.
/// </param>
/// <returns>成功则返回0, 否则返回-1.</returns>
public int SetOutputFormat(SpeechTranscriberRequest request, string value)
{
return(NativeMethods.STsetOutputFormat(request.native_request, value));
}
/// <summary>
/// 设置session id.用于请求异常断开重连时，服务端识别是同一个会话
/// </summary>
/// <param name="request">
/// CreateTranscriberRequest所建立的request对象.
/// </param>
/// <param name="value">
/// value session id 字符串.
/// </param>
/// <returns>成功则返回0, 否则返回-1.</returns>
public int SetSessionId(SpeechTranscriberRequest request, string value)
{
return(NativeMethods.STsetSessionId(request.native_request, value));
}
/// <summary>
/// 销毁实时音频流识别对象.
/// </summary>
/// <param name="request">
/// CreateTranscriberRequest所建立的request对象.
/// </param>
/// <returns></returns>
public void ReleaseTranscriberRequest(SpeechTranscriberRequest request)
{
NativeMethods.NlsReleaseTranscriberRequest(request.native_request);
return;
}
/// <summary>
/// 设置Socket接收超时时间.
/// </summary>
/// <param name="request">
/// CreateTranscriberRequest所建立的request对象.
/// </param>
/// <param name="value">
/// 超时时间.
/// </param>
/// <returns>成功则返回0, 否则返回-1.</returns>
public int SetTimeout(SpeechTranscriberRequest request, int value)
{
return(NativeMethods.STsetTimeout(request.native_request, value));
}
/// <summary>
/// 设置vad阀值. 可选参数, 静音时长超过该阈值会被认为断句.
/// </summary>
/// <param name="request">
/// CreateTranscriberRequest所建立的request对象.
/// </param>
/// <param name="value">
/// vad阀值. 合法参数范围200～2000(ms), 默认值800ms.
/// vad断句与语义断句为互斥关系, 不能同时使用. 调用此设置前, 请将语义断句setSemanticSentenceDetection设置为false.
/// </param>
/// <returns>成功则返回0, 否则返回-1.</returns>
public int SetMaxSentenceSilence(SpeechTranscriberRequest request, int value)
{
return(NativeMethods.STsetMaxSentenceSilence(request.native_request, value));
}
/// <summary>
/// 设置是否使用语义断句.
/// </summary>
/// <param name="request">
/// CreateTranscriberRequest所建立的request对象.
/// </param>
/// <param name="value">
/// 可选参数, 默认false.
/// </param>
/// <returns>成功则返回0, 否则返回-1.</returns>
public int SetSemanticSentenceDetection(SpeechTranscriberRequest request, bool value)
{
return(NativeMethods.STsetSemanticSentenceDetection(request.native_request, value));
}
/// <summary>
/// 设置是否在后处理中执行数字转换.
/// </summary>
/// <param name="request">
/// CreateTranscriberRequest所建立的request对象.
/// </param>
/// <param name="value">
/// 可选参数, 默认false.
/// </param>
/// <returns>成功则返回0, 否则返回-1.</returns>
public int SetInverseTextNormalization(SpeechTranscriberRequest request, bool value)
{
return(NativeMethods.STsetInverseTextNormalization(request.native_request, value));
}
/// <summary>
/// 设置是否在后处理中添加标点.
/// </summary>
/// <param name="request">
/// CreateTranscriberRequest所建立的request对象.
/// </param>
/// <param name="value">
/// 可选参数, 默认false.
/// </param>
/// <returns>成功则返回0, 否则返回-1.</returns>
public int SetPunctuationPrediction(SpeechTranscriberRequest request, bool value)
{
return(NativeMethods.STsetPunctuationPrediction(request.native_request, value));
}
/// <summary>
/// 设置是否返回中间识别结果.
/// </summary>
/// <param name="request">
/// CreateTranscriberRequest所建立的request对象.
/// </param>
/// <param name="value">
/// 可选参数, 默认false.
/// </param>
/// <returns>成功则返回0, 否则返回-1.</returns>
public int SetIntermediateResult(SpeechTranscriberRequest request, bool value)
{
return(NativeMethods.STsetIntermediateResult(request.native_request, value));
}
/// <summary>
/// 设置音频数据采样率.
/// </summary>
/// <param name="request">
/// CreateTranscriberRequest所建立的request对象.
/// </param>
/// <param name="value">
/// 目前支持16000, 8000. 默认是1600.
/// </param>
/// <returns>成功则返回0, 否则返回-1.</returns>
public int SetSampleRate(SpeechTranscriberRequest request, int value)
{
return(NativeMethods.STsetSampleRate(request.native_request, value));
}
/// <summary>
/// 直接关闭实时音频流识别过程.
/// </summary>
/// <param name="request">
/// CreateTranscriberRequest所建立的request对象.
/// </param>
/// <returns>成功则返回0, 否则返回-1.</returns>
public int Cancel(SpeechTranscriberRequest request)
{
return(NativeMethods.STcancel(request.native_request));
}
/// <summary>
/// 发送语音数据.
/// </summary>
/// <param name="request">
/// CreateTranscriberRequest所建立的request对象.
/// </param>
/// <param name="data">
/// 语音数据.
/// </param>
/// <param name="dataSize">
/// 语音数据长度(建议每次100ms左右数据).
/// </param>
/// <param name="type">
/// ENCODER_NONE 表示原始音频进行传递, 建议每次100ms音频数据,支持16K和8K;
/// ENCODER_OPU 表示以定制OPUS压缩后进行传递, 只支持20ms 16K16b1c;
/// ENCODER_OPUS 表示以OPUS压缩后进行传递, 只支持20ms, 支持16K16b1c和8K16b1c.
/// </param>
/// <returns>
/// 成功则返回0, 失败返回-1.
/// 由于音频格式不确定, 传入音频字节数和传出音频字节数, 无法通过比较判断成功与否, 故成功返回0.
/// </returns>
public int SendAudio(SpeechTranscriberRequest request, byte[] data, UInt64 dataSize, EncoderType type)
{
return(NativeMethods.STsendAudio(request.native_request, data, dataSize, (int)type));
}
/// <summary>
/// 设置是否开启nlp服务.
/// </summary>
/// <param name="request">
/// CreateTranscriberRequest所建立的request对象.
/// </param>
/// <param name="value">
/// 可选参数, 默认false.
/// </param>
/// <returns>成功则返回0, 否则返回-1.</returns>
public int SetEnableNlp(SpeechTranscriberRequest request, bool value)
{
return(NativeMethods.STsetEnableNlp(request.native_request, value));
}
