yandex-speech - wrapper for Yandex speech technologies
After reviewing the review of speech recognition engines , I noticed the Yandex API there. And at the weekend I wrote a small wrapper for Node.js for speech recognition in order to find a mat in my phone conversations. Based on the topic on Habré.
I will not spread the list of obscene words, but npm itself is installed by the team.
Sources and examples of use:
github: www.github.com/antirek/yandex-speech
A few details: The
API key was activated within three hours. In the request, he wrote: "you need a test key, there will be 10-20 requests per day." The key specified in the examples is locked. Please request yours : developer.tech.yandex.ru
An example is written in the documentation , similar to a curl request of the type
It turned out that you can send it like this:
Added a check for the size of the sent audio file, not more than 1 MB, because the service swears on larger files.
Also added a wrapper to generate an audio file by text. Unofficial commentary APIon Habré. To both text2speech and speech2text in one bottle.
I hope that someone else will come in handy on the farm. Questions, wishes, criticism by code?
I will not spread the list of obscene words, but npm itself is installed by the team.
npm install yandex-speech
Sources and examples of use:
github: www.github.com/antirek/yandex-speech
A few details: The
API key was activated within three hours. In the request, he wrote: "you need a test key, there will be 10-20 requests per day." The key specified in the examples is locked. Please request yours : developer.tech.yandex.ru
An example is written in the documentation , similar to a curl request of the type
curl -v -4 -H "Content-Type: audio/x-mpeg-3" -d "@examples/data/1.mp3" "asr.yandex.net/asr_xml?key=yandex_developer_key&uuid=12345678123456781234567812345678&topic=general&lang=ru-RU"
It turned out that you can send it like this:
curl -v -4 -F "Content-Type=audio/x-mpeg-3" -F "audio=@examples/data/1.mp3" "asr.yandex.net/asr_xml?key=yandex_developer_key&uuid=12345678123456781234567812345678&topic=general&lang=ru-RU"
Added a check for the size of the sent audio file, not more than 1 MB, because the service swears on larger files.
Also added a wrapper to generate an audio file by text. Unofficial commentary APIon Habré. To both text2speech and speech2text in one bottle.
I hope that someone else will come in handy on the farm. Questions, wishes, criticism by code?