Meet the technology behind our solutions
In VoiceInteraction we believe that speech will be the future platform to access information technology and we intend to be the leading company in this field, basing our action in a continuous search for innovative solutions, to enhance and add value to companies through integration of speech processing technologies, namely speech recognition and synthesis.
Our technologies move in two complementary areas: recognition allows transcribing the speaker’s speech; the synthesis transforms text into speech and the facial animation system enhances the perceived communication capacity. We apply the knowledge developed in the context of intensive R&D activity.
Recognition and synthesis engines can be integrated in applications through proprietary interfaces (accessible in C ++, Java), standard interfaces (MRCP v2) or accessed remotely via Webservices. The engines are independent of the language, and currently Portuguese, Brazilian Portuguese, Spanish (Spanish), Spanish (Latin America), English (American and British), Italian, German and French are supported.
Specifically, in the area of speech recognition, VoiceInteraction developed a system capable of transforming the sound produced by the speaker into a sequence of words – the AUDIMUS.MEDIA. We highlight here the potential of our technology applied to this system:
- Processes audio from pre-existing sources (files) or produced in real time (streaming)
- It can dynamically combine context-free grammars (SRGS) with statistical grammars
- It supports adaptation to the speaker
- It has acoustic preprocessing modules that provide greater robustness to recognition under adverse conditions
- The recognition system is able to support dictionaries >100,000 words while maintaining real-time processing
- There are versions adapted to specific areas: medicine (various specialties), justice, media (television, internet and radio)
- The results have associated confidence measures to allow error recovery
- It is possible to adjust the phonetic transcriptions of dictionary words