Science! By Russell Holly Dec. 13, 2013 4:29 pm
It is now entirely possible that Stephen Hawking will one day not have to apologize for his American accent when he meets someone, thanks to an incredible synthetic voice system called VocalID.
Establishing a synthetic voice for someone who can’t speak is relatively easy today, but there are only so many artificial voices out there. Software generated vocal patterns are repetitive and flat, making it difficult to convey nuances or emotion during a conversation, or even to have a voice that is uniquely yours.
There’s a great deal of work in the human body that happens in order to create a unique voice, which is why it is so difficult to simulate through software alone. But the team at VocalID have figured out the best way to give someone who has lost the ability to speak, partially or completely, a human sounding voice again.
VocalID creates its finished product by combining whatever is left of the target user’s voice with voices from surrogate donor voices to create a personalized synthetic voice. This could be recordings of the user’s voice, or actual samples taken from other people (donors). The surrogate donor’s voice is used to capture the sounds that the user is unable to make on their own, but the source features from the user are applied as the dominant set. This means that while the voice may not 100% belong to the user, the same inflections and tone structure that the original voice was capable of delivering would be present in the synthetic version.
Voice ID is currently accepting applications for voice donors, technical expertise donors, and business expertise donors on their website. They are open to just about any form of contribution, and seem quite open about how they create their synthetic voices. Currently their synthetic voices are only available on Windows systems, but support for Android and iOS is on the way.
The long term applications for something like VoiceID are incredible. Even if you move away from the immediate desire to provide a voice for the voiceless, this technology could be licensed for personalized translations from one language to another and make it possible for someone to deliver something other than a robotic tone.
Now read: Hubble captures water shooting off the south pole of Jupiter’s ice moon Europa
More...
Establishing a synthetic voice for someone who can’t speak is relatively easy today, but there are only so many artificial voices out there. Software generated vocal patterns are repetitive and flat, making it difficult to convey nuances or emotion during a conversation, or even to have a voice that is uniquely yours.
There’s a great deal of work in the human body that happens in order to create a unique voice, which is why it is so difficult to simulate through software alone. But the team at VocalID have figured out the best way to give someone who has lost the ability to speak, partially or completely, a human sounding voice again.
VocalID creates its finished product by combining whatever is left of the target user’s voice with voices from surrogate donor voices to create a personalized synthetic voice. This could be recordings of the user’s voice, or actual samples taken from other people (donors). The surrogate donor’s voice is used to capture the sounds that the user is unable to make on their own, but the source features from the user are applied as the dominant set. This means that while the voice may not 100% belong to the user, the same inflections and tone structure that the original voice was capable of delivering would be present in the synthetic version.
Voice ID is currently accepting applications for voice donors, technical expertise donors, and business expertise donors on their website. They are open to just about any form of contribution, and seem quite open about how they create their synthetic voices. Currently their synthetic voices are only available on Windows systems, but support for Android and iOS is on the way.
The long term applications for something like VoiceID are incredible. Even if you move away from the immediate desire to provide a voice for the voiceless, this technology could be licensed for personalized translations from one language to another and make it possible for someone to deliver something other than a robotic tone.
Now read: Hubble captures water shooting off the south pole of Jupiter’s ice moon Europa
More...
