If you’ve ever tried transcribing a voice recording, you know how tricky that can get. This is because the recording’s quality might not necessarily be the best, and also depending on how fast you type or write, there’s a good chance that you’ll be mashing on the rewind button quite a bit.

However that could change in the future thanks to Baidu’s AI efforts. The Chinese tech giant has recently released a beta for its SwiftScribe platform, which in case wasn’t already obvious, is an app that helps to transcribe voice recordings and uses a neural network to accomplish that. Why, you ask?

This is because with the use of a neural network, it can help to make sense of the speech that is being heard in the recordings, thus giving it context and allowing it to understand the words that come next. Baidu claims that this will provide relatively accurate transcriptions thanks to “thousands of hours” worth of training on various recordings.

However at the moment, the beta for SwiftScribe is limited to 30-50 people which is actually really small, and they’re also only open to transcriptionists. We expect that eventually Baidu will eventually open it up to others, but for now there’s no word on when that will be or how much it will cost.

Filed in General. Read more about AI (Artificial Intelligence) and baidu.

User Comments