One the largest impacts of the Internet is the reduction of communication and geographical barriers. With the web, we’re all becoming part of a global community more than ever before, but we don’t speak the same language, so for now, we’re reliant upon translation services. One of my must-have Google Chrome extensions offers translation, and while it works well, it’s more of point-in-time service that I use as needed with static information. On today’s real-time web, that poses a challenge — one that Microsoft is working on and showing off at its TechFest event.
This video taken by Todd Bishop of TechFlash demonstrates a prototype, real-time voice translation service. Microsoft calls it the “Translating Telephone” but for now, it’s a demo voice over IP service used in conjunction with other Microsoft unified communications solutions. Like Google’s Voice product, Microsoft’s Exchange Server 2010 supports voicemail transcription. But Redmond is taking things one step beyond a standard voice to text service. The prototype shows a voice to translated text, using a conversation between an English-speaking caller and one that speaks German. It’s not perfect just yet, but the accuracy is impressive, as is the speed of the translations which are computer-spoken.
While the hardware powering smartphones of today aren’t the best tools for translation, this type of service could be implemented in a client-server setup — the two phones used for conversation would simply provide the audio, speech and connectivity, while a communications server processes the translations and serves them up. There would of course be a delay while the speech is translated, but it could be manageable. More likely to interfere with adoption is the sound of the voice — we’ve grown accustomed to hearing the actual voice of the person we’re speaking to on the phone. Phones with translation capabilities would make the conversation less personal due to a synthesized voice, but at least we’d be able to understand one another. Besides, I see a whole new market evolving — translation service providers could sell custom voices. I think I’d spring for an Angelina Jolie voice and I probably wouldn’t care too much if the translation wasn’t quite right. Sprechen sie deutsche, Ms. Jolie?
Related content from GigaOM Pro (sub req’d):
- How Speech Technologies Will Transform Mobile Use
- Company Profile: Microsoft (free)
- Company Profile: Google (free)