These Are Our
Services
Advice and Guidance
It costs nothing to talk. We will happily provide some advice and guidance to get you started.
Additional Services
Verbyx can bring decades of experience to help reduce the time and cost of implementing an ASR program. Talk to us about Design & Architecture, Proposal Reviews & Preparation, and system evaluation & Testing.
Technology Licensing
VRX is the Verbyx world class automatic speech recognizer. It supports multiple recognition approaches (statistical model, constrained grammar and keyword/phrase identification. VRX is available on premise or cloud hosted, through a choice of licensing models.
Acoustic Models
Acoustic models are a critical component for any speech system. A poorly matched acoustic model can have a serious detrimental impact on ASR accuracy. The Verbyx team are experts in acoustic model creation and adaptation. Click more for an understanding of the importance of acoustic models.
Language Models
The most effective type of language model will be dictated by the end use application. Verbyx can provide guidance on language model selection and development. The team are also proficient in the phonetic definition of new and unusual words. Click more to learn about language models.
Requirements Analysis
Speech recognition is a complex technology. When defining system requirements it is paramount that you can evaluate and test competing systems on a level playing field. Verbyx can save you time and money by writing or reviewing your requirements and highlighting potential proposal loopholes.
Did You Know?
There are Similarities Between Buying a House Plant and Buying Speech Recognition Technology!
Sure you can buy a house plant by going to the hardware store or garden center, choosing the plant that is within your price range and according to the label, determine its a nice plant for a shaded indoor area in your home.
Like many people you might enjoy the plant for a few weeks before you realize you did not fully understand the care instructions, it was not ideally suited to your home environment and it was withering and restoring it was beyond your gardening skills.
If only you had picked a more suitable plant. If only you had understood how to nurture and care for your plant.
Like plants, there are a variety of speech recognition types; constrained grammar, statistical modeling, key phrase detection, each suited to a different application. And like selecting the right plants, choosing the wrong speech recognition system can leave some users with a solution that does not suit their environment or skill levels.
ASR needs ongoing nurturing and care to achieve the best possible performance.
Requirements Checklist
There are many questions to ask before buying speech recognition technology. Understanding ongoing costs and vendor support are equally as important as the technology and performance choice.
- How can I improve my acoustic and language models
- What will it cost to add support for additional words and sentences
- What were the specifics of the training data used to create the acoustic model
- What were the test environment conditions and circumstances behind your quoted accuracy figures
- Is your solution standards based
- Are there any circumstances that would prohibit me from replacing the ASR with new technology
- Can I create or adapt acoustic and language models without reliance on the speech technology vendor
- Are there any restrictions on dictionary or language model size and complexity
- Do you have any experience with speech recognition in my specific domain