Nuance Dragon Mobile Assistant adds voiceprint startup
The Dragon Mobile Assistant (DMA) app from Nuance Communications is an independent alternative to general personal assistants such as Apple’s Siri, Google‘s voice search assistant, and Microsoft‘s Cortana. DMA leverages Nuance’s speech recognition, language understanding, and text-to-speech running as a cloud service. The app is available free in English on Google Play, supporting Android 2.3 and above.
Nuance’s focus seems to be providing the technology embodied in DMA as an OEM product that can be private-branded, such as its use by Samsung as “S-Voice” on Galaxy smartphones. Nevertheless, Josh Lipe, Director, Dragon Product Solutions, noted in an interview with Speech Strategy News that between 100,000-500,000 units of Nuance’s Dragon Mobile Assistant had been downloaded directly by consumers, a range indicated by the Google Play store. The approach allows Nuance to first try innovations on their Dragon Mobile Assistant before using them in OEM versions (as well as to collect data on how users make requests that can be used to improve the underlying speech recognition and understanding technology). Nuance speech recognition, although not its natural language processing, are used in Apple’s Siri, with Apple developing internal speech recognition capabilities that could replace the Nuance technology.
Some device manufacturers and service providers are sensitive about, in essence, turning over interaction with users to Apple, Microsoft, and Google, making their devices essentially commodities. Nuance offers an alternative.
The major upgrade in the latest version of DMA is a wake-up phrase that is trained to the user’s voice, thus a voiceprint providing insurance that the phone is unlikely to be awakened by anyone else’s voice saying the wake-up phrase. “Hello Dragon” is the default phrase, but users can choose their own wake-up phrase; whatever is chosen is said three times initially to train the system. Nuance recommends a wake-up phrase of at least three syllables.
If the phone is locked and the “attentive” mode is chosen in preferences, the phone can be unlocked and simultaneously be given a command, e.g., “Hello Dragon, ring Ashley on her mobile.” In this mode, there is biometric protection against the phone being awakened by anyone other than the enrolled user. Once the phone is unlocked (by whatever means, including not having it locked in the first place), there is no biometric protection. The phone is simply unlocked, and DMA can be given a command by pressing a virtual button without the wake-up phrase or any further biometric authentication. And, of course, any other apps on the phone can be used. In addition to the protective aspects of the biometric approach, Lipe indicated that the technology reduces the “false acceptance rate,” reducing the chance that an utterance not intended for the smartphone will wake it up.
Lipe noted that DMA is software running under the Android OS, and thus, when enabled, is always listening and drawing current, leading to a lower battery life. Some always-listening solutions, e.g., one provided by Sensory Inc. and used by many device manufacturers, are tailored to specific hardware and draw significantly less current, e.g., by running on a DSP chip in the device.
Another new feature is that Dragon Mobile Assistant now supports landscape mode. Users can dock their device in landscape mode and interact hands-free, providing a better view during the Drive Mode experience.
Also, the new, in-app browser experience is faster and renders more content. One can now interact directly from the web results and select by voice alternative result options directly from the browser.
Other DMA features include flexible, natural control for calling, messaging, marking a calendar, looking up maps and directions, proactive conference-call dialing, and hands-free, eyes-free Drive Mode. As part of its Driver Mode, with a user’s permission, Dragon Mobile Assistant can also access calendar information, and, when a conference call is approaching, Dragon can remind the user via a pop-up window and voice alert, and automatically prompt a dial-in option. The user simply needs to speak to confirm or decline the option. Nuance previously added location sharing and Friend Finder features, which enable users to share their locations with their friends.