search

julius-speech / julius

Open-Source Large Vocabulary Continuous Speech Recognition Engine
BSD 3-Clause "New" or "Revised" License
1.85k stars 300 forks source link

Question Of Julius #137

Open YasuyukiTanaka999 opened 4 years ago

YasuyukiTanaka999 commented 4 years ago

Hi, my name is Tanaka of Aires Electronics Industry.

I am a beginner in voice recognition and Julius. It may be a question you don't care about, Thank you for your understanding.

This time, we have inquired about Regarding the accuracy of Julius's voice recognition. Raspberry Pi, like Alexa We are in the process of building it for voice operations. *Due to the influence of this new type of corona, we have developed a voice recognition technology with the keyword touchless. We are verifying whether we can propose to our products or customers.

Create your own dictionary with Julius I try to read that dictionary, I am in trouble because the voice recognition does not respond well. For example, I have registered the word "aires". Even if it's not the word "aires," I recognize it as Aires.

About this solution Are there any sites or books that you can refer to?

I'm sorry I'm busy Thank you.

lpawlaczyk commented 4 years ago

Hi. Which language are you trying to add to your solution? The best way to add words to recognition is to use grammar. In case of normal speech recognition with language model it is more difficult as the word you are trying to add should be a part of language corpus.

YasuyukiTanaka999 commented 4 years ago

Thank you very much for contacting me.

The language is Japanese.

It means using "Grammer-Kit" instead of "dictation-kit".

I understand. I will try.

If you have any questions, please let me know.

Thank you

From: Lesław Pawlaczyk [mailto:notifications@github.com] Sent: Monday, July 27, 2020 5:07 PM To: julius-speech/julius julius@noreply.github.com Cc: YasuyukiTanaka999 y-tanaka@ailes.co.jp; Author author@noreply.github.com Subject: Re: [julius-speech/julius] Question Of Julius (#137)

Hi. Which language are you trying to add to your solution? The best way to add words to recognition is to use grammar. In case of normal speech recognition with language model it is more difficult as the word you are trying to add should be a part of language corpus.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/julius-speech/julius/issues/137#issuecomment-664188557 , or unsubscribe https://github.com/notifications/unsubscribe-auth/AQNGIV6N6IETT2ZVMDSYQBTR5UYSZANCNFSM4PIMCPNA .