Join GamesBeat Summit 2021 April 28-29. Register for free or get a VIP card with a discount today.
Google currently has an open source from Lyra in beta, an audio codec that uses machine learning to produce high quality voice calls. The code and demo, available on GitHub, compresses raw audio down to 3 kilobits per second for “quality that compares favorably with other codecs,” says Google.
While mobile connectivity has been steadily increasing over the last decade, the explosive growth of computing power on the device has surpassed access to reliable, fast internet. Even in areas with reliable connections, the emergence of work from anywhere and teleworking has stretched the data line. For example, early in the pandemic, nearly 90 of the 200 largest cities in the United States saw Internet speeds decline as bandwidth was strained, according to BroadbandNow.
Lyra’s architecture is divided into two pieces, an encoder and a decoder. When someone speaks into the phone, the encoder captures characteristic attributes, called features, from their speech. Lyra extracts these functions in 40 millisecond chunks and compresses them and sends them over the network. It is the decoder’s job to convert the functions back to a sound waveform that can be played over the listener’s telephone.
According to Google, Lyra’s architecture is similar to traditional audio codecs, which form the backbone of Internet communication. But while these traditional codes are based on digital signal processing techniques, the main advantage for Lyra comes from the decoder’s ability to reconstruct a high quality signal.
Google believes there are a number of applications Lyra can be unique to, from archiving large amounts of voice and saving battery to relieving network load in emergencies.
“We are excited to see the creativity that the open source community is known for using on Lyra to come up with even more unique and effective applications,” Google Chrome engineers Andrew Storus and Michael Chinen wrote in a blog post. “We [want] to activate developers and get feedback as soon as possible. ”
The Lyra code is written in C ++ using the Bazel build framework. The core API provides an interface for encoding and decoding at the file and packet level, and the complete signal processing tool is provided, which includes filters as well as transformations. Google’s sample code integrates with Android NDK to show how Lyra can work with Java-based Android apps, and Google has also provided weights and vector quantizers needed to run Lyra.
“This release provides the tools needed for developers to encode and decode audio with Lyra, optimized for the 64-bit ARM android platform, with development on Linux,” Storus and Chinen continued. “We hope to expand this code base and develop improvements and support for more platforms in line with society.”
VentureBeat’s mission is to be a digital marketplace for technical decision makers to gain knowledge about transformative technology and transactions. Our website provides important information about computer technologies and strategies to guide you as you lead your organizations. We invite you to become a member of our community, to gain access to:
- updated information on the topics you are interested in
- our newsletters
- gated thought leader content and reduced access to our valued events, such as Transform 2021: Learn more
- network features and more
Become a member