Real-time speech-to-speech translation on the menu for Google
Google recently announced that it was working on a real-time speech recognition app for phones powered by the Android operating system. The Internet giant has a strong interest in language translation technology and the last six months alone have seen Google launch automatic captions (autocaps) for YouTube videos which use automatic speech recognition (ASR) technology; the inclusion of new languages and additional features with Google Translate; and there was its collaboration with Wikipedia in relation to the Swahili language translation competition. Declaring that this app could be market-ready within the decade has caused a few eyebrows to be raised in the linguistic community. However, by going to press just before this week’s Mobile World Congress 2010, meant the announcement sent ripples of excitement through the mobile technology and cell phone market.
Google already has the technology in place to allow for speech-to-text translation, with both the Nexus One phone and also the Google Maps application where audio input is converted into text. However, these automatic real-time translation developments are significant because, as reported in The Times (7 February 2010), the new software will be ‘capable of understanding a caller’s voice and translating it into a synthetic equivalent in a foreign language.’ Speech will be analysed in ‘packages’ and the phone will be intelligently trained to recognize the speaker’s voice. Exciting though these developments are, there is always the caveat emptor which comes hand in hand with automatic translation tools, in that mistranslations could arise due to the lack of human post-editing. Professional translators and interpreters study for many years to perfect their skills and the pitfalls of automatic machine translation could be heightened if the phone is unfamiliar with a speaker’s accent, or if the audio input becomes distorted by background noise for example.
Living in the digital world is no guarantee that people in the street will be willing to provide their voices for this intelligent translation technology software. Nevertheless, the ability of such a device to gain basic information in emergency situations has not gone unnoticed. Another Google language product is the GPS navigation system with voice guidance for phones with Android – and if combined with real-time two-way translation software, such a device could prove to be extremely useful for military, medical and humanitarian personnel.
Meanwhile, at the Mobile World Congress 2010, Toshiba will be demonstrating its TG02 smartphone with Voice Translation app. The technology news website Ixplora.com reports that speech recognition, machine translation and speech synthesis technology translates the speaker’s input language in real-time, producing a synthesized audio output in the chosen foreign language – all without the need for an Internet connection.
The World Mobile Congress is being held until 18 February in Barcelona where it will be seen whether the showcased technologies meet the approval of technophiles worldwide.