Technology
The age of the cellphone has arrived, and with it the era of the speech interface. Promptu has mastered the technology to put a world of information and applications at the user's disposal, all controlled by spoken command.
Accuracy
The heart of a successful speech interface is high accuracy. Promptu's innovative architecture, proprietary speech data base and processing algorithms, expert speech technologists and UI designers, and in-field experience combine to ensure the highest possible accuracy.
Architecture
Promptu pioneered the company's industry-leading Network Speech Recognition (NSR) architecture. NSR divides speech recognition between a handset-resident client, which captures the audio to be recognized, and a remote server, with computational resources far exceeding those of the phone.
Audio captured by the client is sent to the remote recognition server over the mobile data channel. No conventional telephone call is involved. Since the data travel over an error-controlled, flow-controlled TCP/IP connection, the audio reaching the server is crystal-clear: there is literally no channel noise, and there are no audio dropouts.
At the receiving end, a powerful Promptu server analyzes the user's utterance, recognizes and interprets its content, and returns the desired information in a flash.
Languages
Promptu's servers operate in English, Italian, French, German, and Mandarin. The company's technology is highly adaptable and, with suitable acoustic and textual databases, can be extended to any human language.
Promptu's products operate with both grammar-based and open-dictation recognizers, as the application requires. For travel and transportation products, where the task is recognizing a spoken city, station name, street address, or the like, grammar-based recognition is more appropriate. For messaging products, like SMS message creation, open dictation is used.
Security, Reliability, Scalability
Promptu has built and deployed systems that meet the standards of some of the world's largest corporations, including mobile telephone operators. The company's products meet industry-standard SLAs, and are self-monitoring, fault-tolerant and self-healing: if a given hardware or software component fails, the system automatically reconfigures itself for continued operation without the failed component, then reports the outage to Promptu's 24/7 operations staff.
This same technology allows for zero-downtime system upgrades: additional compute power can be added dynamically to a production server, which will automatically detect the new capacity, and exploit it for higher throughput.
Data and Insight
As Promptu's products operate, they gather ever-larger repositories of anonymized acoustic and textual data. These data are used to periodically regenerate Promptu's speech recognition systems, increasing recognition accuracy through expanded coverage of accents and speech patterns. At the same time, Promptu's speech and UI design experts engage in ongoing technical and market research, continually refining Promptu's product designs and speech processing algorithms.
Patents
Promptu's patent portfolio includes 38 issued or pending patents, both worldwide and domestic. The patents span a wide range of subject matter, from Promptu's innovative applications, to elements of the company's core speech and language technology.