Convert audio to text
-
15-04-2021 - |
Question
I just want to know if there is any build in libraries or external libraries in Java or C# that allow me to take an audio file and parse it and extract the text from it.
I need to make an application to do so, but I don't know from where I can start.
Solution
Here are some of your options:
OTHER TIPS
Here is a complete example using C# and System.Speech
The code can be divided into 2 main parts:
configuring the SpeechRecognitionEngine object (and its required elements) handling the SpeechRecognized and SpeechHypothesized events.
Step 1: Configuring the SpeechRecognitionEngine
_speechRecognitionEngine = new SpeechRecognitionEngine();
_speechRecognitionEngine.SetInputToDefaultAudioDevice();
_dictationGrammar = new DictationGrammar();
_speechRecognitionEngine.LoadGrammar(_dictationGrammar);
_speechRecognitionEngine.RecognizeAsync(RecognizeMode.Multiple);
At this point your object is ready to start transcribing audio from the microphone. You need to handle some events though, in order to actually get access to the results.
Step 2: Handling the SpeechRecognitionEngine Events
_speechRecognitionEngine.SpeechRecognized -= new EventHandler(SpeechRecognized); _speechRecognitionEngine.SpeechHypothesized -= new EventHandler(SpeechHypothesizing);
_speechRecognitionEngine.SpeechRecognized += new EventHandler(SpeechRecognized); _speechRecognitionEngine.SpeechHypothesized += new EventHandler(SpeechHypothesizing);
private void SpeechHypothesizing(object sender, SpeechHypothesizedEventArgs e) { ///real-time results from the engine string realTimeResults = e.Result.Text; }
private void SpeechRecognized(object sender, SpeechRecognizedEventArgs e) { ///final answer from the engine string finalAnswer = e.Result.Text; }
That’s it. If you want to use a pre-recorded .wav file instead of a microphone, you would use
_speechRecognitionEngine.SetInputToWaveFile(pathToTargetWavFile);
instead of
_speechRecognitionEngine.SetInputToDefaultAudioDevice();
There are a bunch of different options in these classes and they are worth exploring in more detail.
You might check Microsoft Speech API. I think they provide a SDK that you can use for your objective.
For Java, it seems there is a solution from Sun: javax.speech.recognition
You can use SoX (the Swiss Army knife of sound processing programs) to convert audio file to text file with numeric values corresponding to sound frequency/volume.
I have done it for a previous project but don't know the exact command options.
Here is a link to the project: http://sox.sourceforge.net/Main/HomePage