Social Media and Speech Recognition
Custom Solutions for Social Media
We Take a Backseat to Your Product
MatrixHCI is in the business of delivering speech solutions that make you look good. That means that our development and our product take a backseat and quietly deliver speech recognition results from within your product line.
For all intents and purposes, your customers will never know we exist.
When Out-of-the-Box Speech Recognition Does Not Work
There are many prepackaged speech recognition applications designed to perform very specific tasks like legal dictation, medical dictation, etc. These products have been specifically designed for a single dictation process in very narrow fields where language and acoustic models can be designed to meet a majority of the speech results for very narrow vertical markets.
Additionally, although free, Siri and Google speech recognition solutions have very narrow applications and are, in most cases, tied to the vendor’s own specific software applications and they are not customizable or usable by outside vendors for integration into their own software solutions.
In a world where social media applications are growing exponentially and where many of the applications are distributed free or for a nominal fee, these products have severe and cost prohibitive limitations. Although these products work well in their intended environment they have several major drawbacks when you need more than a prepackaged solution that needs to be customized and/or integrated into your specific solution.
As an example, these products:
- Are very narrow in their application solutions.
- They are not open for customization or integration into your specific applications.
- No specific custom acoustic or language models can be developed by the supplier.
- No source code is available and no specific application interfaces (APIs) are available for developers to integrate such products.
- There is no way to modify the acoustic models and language models for languages not offered.
- They do not offer any trainable artificial intelligence or dialogue management capabilities.
- They are also cost prohibitive if you need to bundle the stand alone app with your product.
MatrixHCI offers custom development of speech solutions that are specifically designed to meet your unique speech needs. The licensing can be configured as a per use or a one-time fee so you can maintain value proposition when your overall product has a free distribution model.
Let MatrixHCI consult with you to determine the best approach for meeting your specific speech recognition needs. We can design a speech solution that will custom fit your desired application requirements in any language and in any acoustic environment.
Client / Server Design
Because the core of our transcription system is based on a client server architecture, the actual speech recognition engine can be housed in the Cloud where high powered MatrixHCI designed servers can process the audio and return transcribed results to your client software. This is important when you need the higher horsepower that cannot be found in smaller handheld devices or slower computers to handle the high volume of speech recognition processing.
In this case, a small thin client software component can exist on your customer client (on a smart phone, handheld device, or PC) which packages the audio collected and takes care of all management to send the audio to the server and return the results … all the time using the higher horsepower servers to do the “heavy lifting” needed for high volume real-time speech recognition.
This client can be integrated behind the scene in the back end of your client software and it can handle all the complexities of sending and receiving the audio to the server. This architectural design is much the same as that which is employed by Siri and Google where the actual speech recognition does not reside on the consumer device.
The main difference between us and Siri and Goggle is that our technology is totally customizable and is not controlled or limited to specific devices and their own applications. MatrixHCI will design the architecture using your unique needs and interface requirements and sit within your application thus speech enabling your application in a way that meets your needs.
Additionally, when the devices your applications run on have the horsepower and hardware real estate to handle the speech recognition “on device,” the MatrixHCI architectural design allows the server to actually reside on the device itself thus eliminating the need for an internet connection. This means that you will never get the “Siri Not Available. No Internet Connection” message on our custom designed product.
