Deep Neural Networks (DNN) for speech model training
DynaSpeak® is a speech recognition platform designed for maximum versatility. As the quality of speech to text dictation has improved, use of speech-based user interfaces has evolved from rudimentary phone menu systems to cutting-edge interfaces in mobile, automotive, consumer electronics, and entertainment applications. DynaSpeak gives developers the flexibility to customize DynaSpeak for their unique domain and deployment environments.
SRI provides the DynaSpeak TryIt! web service for free, non-commercial trial use. Developers who use SRI’s BuildIt! web API may use the DynaSpeak web service on a freemium basis, depending upon usage. See usage options for more information. In addition to the DynaSpeak web service, SRI provides the DynaSpeak Development Platform on a standalone basis under separate licensing terms. Contact us and tell us more about your product or service needs and we can tailor a solution for you.
Uses DNN for speech model training to achieve state of the art speech recognition performance
Supports use in Internet-connected services or on a standalone basis with no network connection
Scales from mobile devices and embedded processors to server-class cloud systems
Includes toolsets that allow developers to tailor their applications for varied domains and languages
Gives developers ownership, retention, and control of their customers' speech interaction data
Start by doing what's necessary; then do what's possible; and suddenly you are doing the impossible.
Francis of Assisi
After successful sign-up and/or login, please browse to the settings page to obtain the API key, which is unique to your developer account.
Prepare an audio file which contains speech that needs to be recognized as text. A wide range of audio file formats are supported. Please visit here to learn all the audio formats accepted.
Upload the file by POSTing to https://api.sri.com/v1/dynaspeak_now as form data (e.g., file type input) together with the API key in the header. Parameter name for the file should be file and the header name for the API key should be x-api-key. File audio format will be auto-detected by the API service. Below is the curl example with the file name and the API key to be replaced with your own values.
curl -F "file=@<local path to audio file>" -H "x-api-key: <API key>" https://api.sri.com/v1/dynaspeak_now
Upon successful processing of the file, following format JSON will be returned from the server.
{
"type": "rendered"
"status": 200
"message": "Hello World"
}
For full details about DynaSpeak API, please visit Documentation
You can use DynaSpeak APIs free of charge up to 1,000 minutes of audio. If you would like further information, please contact us at platform-support@sri.com.
Keep in touch with us