Mobile Speech Technology
NowSpeak Technologies offers the latest text-to-speech and speech recognition technologies specifically targeted and optimised for low resource mobile devices. These technologies can be used independently or combined to provide completely hands busy, eyes busy control of devices.
Whatever type of voice interaction you require NowSpeak has the technology and expertise to provide a cost effective and reliable solution.
Text-to-Speech
The ability to synthesise speech from text in real time means it is possible to keep a user informed even if your device does not have a display or when the user is not able to read the display.
There are internationally recognised safety and legal issues with using mobile devices while driving. Using speech output in these circumstances provides clear benefits, allowing the user to stayed informed even when they their hands and eyes maybe busy.
For instance, text-to-speech can be used to inform user of changes to a device's status, tell the user who is calling or what music track is being played and speak text messages such as SMS and emails.
Text-to-speech core features:
• Natural, smooth and intelligible speech
• All major languages supported
• Rapid development of new languages and voices
• Small memory footprint per language
• Cost effective to have multiple languages in a device
• System optimised for embedded applications
• Support for intonation
• Choice of voices available
• Customisation of voices
Speech Recognition
We use the latest speaker independent technology that works out of the box with no user training or registration required. It supports large vocabularies and can be used to recognise words, digits or whole sentences.
Our systems are trained to recognise even when people use different pronunciations or have accents. Our speech recognition is designed to work in real world conditions.
We can also supply speaker dependent modules. These require training by the user who must provide a voice tag, a word or phrase, that is stored by the speech engine and associated to a contact name or command.
Speech recognition core features:
• Speaker independent continuous speech recognition
• Built-in noise robustness
• Low memory and CPU footprint which can be scaled to suit platform
• Adaptation to user and environment
• Speaker dependant module available
• Separate phonetic and digit models supported
• Arbitrarily complex finite state grammars
• Garbage modelling provides rejection of invalid commands and noise
• Recognition results as N-best list and confidence scores
NowSpeak speech recognition and text-to-speech engines have been designed to make them easy to put onto any hardware platform with or without an operating system.
We provide support for the following platforms:
• CSR BlueCore
• ARM
• Symbian
• Linux
• Windows Smartphone
• Windows PocketPC
• Palm