Home / FAQ's

Frequently Asked Questions FAQ's

 

Feel free to email or call, we would love to hear from you. In the meantime, here are answers to some of the questions we encounter most often.

How accurate is your speech recognition and can you provide benchmarks against your competitors?

In all side by side tests performed to date, VRX out-performs competing offerings. We do not offer benchmark data as the results can often be misleading. Accuracy comparisons are only useful if done side-by-side and in identical cirumstances with matching hardware resources and language models. The test audio must be clean (i.e., not available to the vendor prior to testing).

Be very wary of companies boasting a (insert very high number here) % accuracy. It is vital that your specific environment is fully understood before any indication of performance can be given. Any numbers given prior to such discussions should be treated with caution.

We are confident that VRX will not be beaten in accuracy. Verbyx will fully refund your licence fee if you are not satisfied with the performance.

What operating systems do you support?

VRX will operate happily on most versions of Microsft Windows and in 32bit or 64bit modes. We can also offer support for Linux and other operating systems are available on request.

How robust is your software?

VRX is a modern industry grade ASR with robust resource and thread management (no leaks, no data-races, no dead-locks). VRX is warrantied against defects.

What languages do you support?

Verbyx has developed a unique capability for developing new acoustic models with limited training data. We can very quickly support new languages as requested. We could happily provide a long list of supported languages but we believe such lists are often misleading. Acoustic models need to be matched with the audio environment of the end user application. Contact us with specific requirements.

Do you support cloud applications?

Yes. VRX can be used in a cloud environment. If you need help in developing a cloud deployment, we can offer services in this area.

Can you provide a demonstration system?

Providing a demonstration system is a possibility but only under agreed conditions. We generally do not provide test and demonstration systems for free play. If you have competence in developing language models and you can outline how you plan to assess performance of VRX then we would be happy to discuss availability of demonstration software.

What modes of operation does VRX support?

VRX currently supports 3 modes of operation, statistical language modeling, context free grammar and key phrase detection. With these modes, most application types can be addressed, i.e., transcription, command and control, language assesment, voice analytics and many more.

What about Natural Language Processing (NLP)?

NLP is a field of computer science, artificial intelligence, and linguistics concerned with the interactions between computers and human (natural) languages. The input for the NLP is typically the output from speech recognition It is frequently used in speech enabled applications such as the latest crop of personal digital assistants found on each of the major mobile operating systems. NLP is in itself a complex task that requires specialist knowledge and detailed application specific preparation. Verbyx prefers to concentrate on the input to NLP systems (speech recognition) and is happy to support the integration of best in breed NLP.