Azure认知语音服务STT -部分文本

Question

在我的代码(下面)中，当我通过STT处理它时，它只给出整个音频的第一个字母/单词。

音频有"A、B、C、D、E、F“

我遗漏了什么？

Imports Microsoft.CognitiveServices.Speech
Imports Microsoft.CognitiveServices.Speech.SpeechConfig
Imports Microsoft.CognitiveServices.Speech.Audio

Module Module1

    Sub Main()
        Dim SpeechConfig As SpeechConfig = FromSubscription("<CHANGED>", "eastus")
        Dim audioConfig As Audio.AudioConfig = Audio.AudioConfig.FromWavFileInput("<CHANGED>.wav")
        SpeechConfig.OutputFormat = Microsoft.CognitiveServices.Speech.OutputFormat.Detailed
        Dim recognizer As New SpeechRecognizer(SpeechConfig, audioConfig)
        Dim result = recognizer.RecognizeOnceAsync().Result

        Select Case result.Reason
            Case ResultReason.RecognizedSpeech
                Console.WriteLine($"RECOGNIZED: Text={result.Text}")
                Console.WriteLine($"    Intent not recognized.")
            Case ResultReason.NoMatch
                Console.WriteLine($"NOMATCH: Speech could not be recognized.")
            Case ResultReason.Canceled
                Dim cancellation = CancellationDetails.FromResult(result)
                Console.WriteLine($"CANCELED: Reason={cancellation.Reason}")

                If cancellation.Reason = CancellationReason.[Error] Then
                    Console.WriteLine($"CANCELED: ErrorCode={cancellation.ErrorCode}")
                    Console.WriteLine($"CANCELED: ErrorDetails={cancellation.ErrorDetails}")
                    Console.WriteLine($"CANCELED: Did you update the subscription info?")
                End If
        End Select

    End Sub

End Module

您可以在github这里下载音频文件，https://github.com/ullfindsmit/StackOverflowAssets/blob/master/abcdef.wav

另外，如果你知道我在哪里可以得到更详细的STT数据，我会很感激的。我正在寻找的是一个JSON输出，它表示开始时间和结束时间以及单词和/或句子。

非常感谢你的帮助。

更新，因此异步处理程序由于某种原因不能为我工作，但是下面的代码确实起作用了

        While True
            Dim result = recognizer.RecognizeOnceAsync().Result
            Select Case result.Reason
                Case ResultReason.RecognizedSpeech
                    Console.WriteLine($"RECOGNIZED: Text={result.Text}")
                    Console.WriteLine($"    Intent not recognized.")
                Case ResultReason.NoMatch
                    Console.WriteLine($"NOMATCH: Speech could not be recognized.")
                Case ResultReason.Canceled
                    Dim cancellation = CancellationDetails.FromResult(result)
                    Console.WriteLine($"CANCELED: Reason={cancellation.Reason}")

                    If cancellation.Reason = CancellationReason.[Error] Then
                        Console.WriteLine($"CANCELED: ErrorCode={cancellation.ErrorCode}")
                        Console.WriteLine($"CANCELED: ErrorDetails={cancellation.ErrorDetails}")
                        Console.WriteLine($"CANCELED: Did you update the subscription info?")
                    End If

                    Exit While
            End Select
        End While

Answer 1

RecognizeOnceAsync方法只识别“一次”.音频数据文件中包含的第一个“话语/短语”。如果你想识别多个短语，你可以做以下两件事中的一件：

1. 反复给RecognizeOnceAsync打电话..。在识别最后一个短语之后，对该方法的下一次调用将返回一个将Canceled.
2. Switch设置为result.Reason的结果，从使用RecognizeOnceAsync到使用StartContinuousRecognitionAsync，并将事件提交器挂钩到Recognizing事件。事件回调将允许您通过检查传递的SpeechRecognitionEventArgs来查看结果，如：e.Result .

通过像这样运行Speech CLI，您可以看到这两种行为：

spx recognize --once+ --key YOUR-KEY --region YOUR-REGION --file "https://github.com/ullfindsmit/StackOverflowAssets/blob/master/abcdef.wav"
spx recognize --continuous --key YOUR-KEY --region YOUR-REGION --file "https://github.com/ullfindsmit/StackOverflowAssets/blob/master/abcdef.wav"

您可以在这里下载演讲CLI：https://aka.ms/speech/spx-zips.zip