The development of speech-recognition interfaces may finally unleash the versatility of mobile phones. While mobile phones can do everything from downloading music to browsing the Web, the practicality of these tasks has been limited. Most people avoid anything beyond calls and short text messages because of the tediousness of the data entry required by more complicated functions. Hoping to fill this gap is Vlingo, a startup about to market a simple speech-recognition interface for mobile phones. Vlingo was founded by Michael Phillips and John Nguyen, a founder and computer scientist, respectively, of SpeechWorks, since acquired by Nuance Communications, which provided call-center speech interfaces for its clients. Vlingo's innovation lies in its ability to convert speech into text, and then allow users to correct errors via navigation keys. More impressively, these user-generated corrections help the system to learn. Because the spoken words are processed at Vlingo's server, the system can deliver written transcriptions in less than two seconds. A test search for the phrase 'Schumann Piano Concerto' returned the second and third words intact. A manual entry of 'Schumann' was needed, but having been entered once, that term will likely reappear as a clickable option for words that sound like 'Schumann' or for music-related searches. According to CEO David Grannan, Vlingo hopes to have carriers bundle its interface with their offerings, on the assumption that user searches will generate revenue for them. Vlingo is not alone, however. Mazin Gilbert of AT&T Labs says that others, including AT&T, are developing speech interfaces for mobile phones.