-5

I would like to convert spoken words into text so I can use nslinguistictagger in my app. How can I convert speech to text? What are the options? Does openears support voice to text conversion?

tugce
  • 651
  • 3
  • 13
  • 28

2 Answers2

4

OpenEars developer here. Yes, OpenEars does speech recognition and text-to-speech. You need to define a language model or grammar for it containing your vocabulary, but it can be done automatically from an NSArray of word or phrase NSStrings, or a text corpus.

Halle
  • 3,584
  • 1
  • 37
  • 53
  • 1
    I'm new to this speech recognition area, forgive me for my stupid question. But rather than text-to-speech I need speech-to-text as dictation, does openEar help me on this? – tugce Mar 25 '13 at 09:05
  • see my answer... I clearly mentioned about OpenEars TTS. – Shamsudheen TK Mar 25 '13 at 09:06
  • Hi Ramshad, good answer. The OP is asking about speech recognition rather than TTS in this case. @tugce, speech recognition is the more common term for speech-to-text, so when I explained that OpenEars does both speech recognition and text-to-speech I meant to reassure you that you could do speech-to-text with it. – Halle Mar 25 '13 at 09:47
  • @Halle could you please guide me on where to look for speech-to-text by using OpenEars – tugce Mar 25 '13 at 09:51
  • http://www.politepix.com/openears and http://www.politepix.com/openears/tutorial – Halle Mar 25 '13 at 09:53
  • I have already checked out there, but as @Ramshad stated openears seems to me only support specific words recoginiton, please correct me if I'm wrong. I need to dictate(recognize) whole speech not the specific words. – tugce Mar 25 '13 at 09:54
  • If you want to do what's called large vocabulary recognition and attempt to recognize any word that the user might say, you have to use a server-based API rather than an offline API. Public APIs include AT&T Watson, Nuance Dragon SDK, and iSpeech, all with different pricing structures -- just Google any of them to get started. – Halle Mar 25 '13 at 09:56
  • This is getting really far away from the topic, but crossing an inside developer is a good chance. http://stackoverflow.com/questions/15527107/speech-recogition-and-intonation-detection Let me explain what's in my mind and I would be glad if you comment on whether I could do it with OpenEars or not- I would definitely want make use of it. Counting number of questions in a speech, so I can easily count wh questions with recognition. Moreover, look for intonation change or nslinguistictagger, that is why I need dictation. – tugce Mar 25 '13 at 10:04
  • @Halle: he is mentioned about detecting whole sentence. I think Openears do well this one.Only he need to do is, create vocabulary files with long sentences.I know ,You have clearly mentioned it in ur website. tugce: create a .dic and .language model files with your sentences. it will detect the long sentence then.btwn please read the Openears documents well.. – Shamsudheen TK Mar 25 '13 at 10:06
  • The sentences to be spoken are not pre-defined, so I cannot create models,that's why I need dictation with large vocabulary recognition. – tugce Mar 25 '13 at 10:17
  • Hi Ramshad, I agree with you that it would work quite well with advance knowledge of the sentence content because the time to generate the model would be trivial. It sounds like tugce doesn't have access to that info, so it will probably have to be server-based. – Halle Mar 25 '13 at 10:21
  • 1
    @Halle: yup. yu]ou are absolutely correct.btwn thumbs-up for your great Openers library,,, – Shamsudheen TK Mar 25 '13 at 16:39
  • @Halle thanks for the library and answer – tugce Mar 27 '13 at 15:17
2

Openears will support free speech recognition and text-to-speech functionalities in offline mode.

They have FliteController Class Reference, which controls speech synthesis (TTS) in OpenEars.

They have done an excellent job in speech recognition area.

However, please note that it will detect only the words that you mentioned in vocabulary files.It iss good to work as offline mode to get the better performance.

@Halle: Correct me if I'm wrong.

You have a paid option, Dragon Dictation which is working as online engine.

or use VocalKit: Shim for Speech Recognition on iPhone

I would like to point out that , none of them are accurate than Siri (Siri SDK is not available yet).

Shamsudheen TK
  • 30,739
  • 9
  • 69
  • 102