Google Voice-to-Voice

The Google Voice-to-Voice application was conceptualized as part of the two day Google Case Challenge held by the Canadian Undergraduate Technology Conference (CUTC) in 2010. The case challenge was to develop methods in monetizing Android, at the time, the new upstart OS opposed to Apple's and RIM's offerings.

Our team focused on Google's international reach, and the idea of connecting people together using Android. We wanted to give people a medium to collaborate and communicate regardless of location, culture, or language. To minimize development costs and time, we opted to use three of Google's existing (at the time) services: Google "search by voice", Google Translate, and Google Voice.

The Concept

The overall concept consisted of five stages:

1. Voice Streaming - speech from calls on Android phones would be streamed to a remote Google server.

2. Voice-to-text - using software from Google's "search by voice" service, the server would transcribe the speech into regular text.

3. Text-to-text - using software from Google Translate, the server would translate the text into another language (chosen by the original speaker)

4. Text-to-voice - using software from Google Voice, the server then takes the translated text and returns it to speech.

5. Voice streaming - finally, the translated speech is streamed to the listener's Android phone.

The Interface

Shown here on a screenshot of the HTC Hero, Google Voice-to-Voice (V2V) is available from the calling screen. Once a call has been established, the user can simply tap the translate bar (seen above the "End call" button), which will slide out a language selection tab. The user then chooses the language they are speaking, and the language they wish to translate to, and the rest of the process is carried forward on Google's servers.

Personalization and User Experience

Speech is highly personalized to each individual, so a system would be developed to store a user's speech patterns on Google's servers in order to improve the speech recognition software over time. This may also allow the system to read back translated text in the original user's own voice, aiding in natural sounding conversations. Users also had the option of saving preset translation languages for specified contacts.

Additional Features

The same concepts could also be applied to SMS (texting), and be potentially less complicated, with the translation done on the Android phone instead of Google's servers.