Inside Google & Samsung’s Open Source Spatial Audio Project w/ Eclipsa Audio | Ep 36

Three Things You Can Take Away From This Episode

• Eclipsa Audio is fully open source and royalty-free, so anyone can build encoders, players, plugins or services on it without asking permission.
• Think immersive from day one. A spatial mix done as an afterthought will never match one planned from the start.
• One Eclipsa file can carry a dedicated stereo mix, an immersive mix and alternate languages, so the listener always gets the version made for their system.

Immersive audio has been around for decades, but it has mostly stayed behind licence fees, specialist hardware and a small audience of enthusiasts. Eclipsa Audio sets out to change that. It started as a collaboration between Google and Samsung, and it’s now an open, royalty-free format that YouTube has been streaming since January 2025.

In this episode, Josh is joined by three guests who see the format from very different angles. Jani Huopaniemi leads the Eclipsa Audio programme at Google. Scott Kramer was the first sound expert hired by the Netflix studio and now runs his consultancy, Greenfield Sound. Klaus Trelby is an engineer and producer who has spent more than 20 years mixing immersive music, including the 5.1 mix of Fleetwood Mac’s Rumours.

Between them, they cover what the format is, how creators can start using it today, and what it could mean for music, film, TV and developers.

1. What Eclipsa Audio Actually Is

The underlying standard is IAMF (Immersive Audio Model and Formats), developed in the Alliance for Open Media alongside members such as Netflix, Meta and several universities. Eclipsa Audio is the consumer-facing brand, because IAMF is, as Jani puts it, a mouthful.

IAMF is a container format. It defines how to put multichannel immersive audio into a file, currently MP4, with recommendations for rendering to loudspeakers and binaural headphones. The specification is royalty-free, and the encoders, decoders and tools are all on GitHub for anyone to use.

2. How It Differs From Dolby Atmos

Beyond being open source, Eclipsa streams what you mix. A 7.1.4 mix is delivered as 7.1.4, and the playback device adapts it to whatever system it has. Creators can also combine channel-based beds with ambisonics in the same file, and compression uses Opus, the well-established open source codec.

Audio objects with positional metadata aren’t in the current 1.0 and 1.1 specifications, but they’re expected in future versions. Klaus points out that streamed Atmos reaches listeners as a channel-based mix anyway, so the listening experience is closer than it might seem.

3. Getting Started as a Creator

The quickest way in is the free Eclipsa Audio plugin, available as AAX for Pro Tools and as VST3, tested in Reaper, Logic Pro and Premiere Pro. Jani notes that a VST3 build doesn’t automatically work in every DAW, so support is being tested host by host.

Creators who prefer the command line can use FFmpeg, which already supports IAMF. Third-party tools are arriving too, including batch converters that turn an existing Atmos catalogue into Eclipsa. The team is also working on mobile capture, so you could record spatial audio on a phone and upload it straight to YouTube.

4. Why Immersive Music Stalled Before

Klaus has watched immersive music struggle for years. DVD-Audio and Super Audio CD were great formats, but they only reached the one or two percent of people with a home theatre, and there was no binaural or fold-down option. When labels stopped seeing a return, the formats died.

Atmos brought immersive music back, but as a licensed format everyone has to pay to use, from hardware makers to engineers. He also thinks the name has lost its meaning for consumers, much like “Xerox”. What excites him about Eclipsa is the idea of someone buying a new TV and soundbar, opening YouTube, and suddenly hearing their speakers do something they’ve never done before.

5. Think Immersive From Day One

Klaus sees the same mistake again and again. Artists focus on the stereo mix, and then at the finish line someone asks about an immersive version. It becomes a rushed job, or worse, the label asks for stereo stems and makes the immersive mix itself.

His advice is to plan for it from the start. Record an extra guitar or vocal, because it’s easier to leave parts out of a stereo mix than to add them later. You don’t need a studio like his. Start binaurally on headphones, get inspired on a simple 5.1 system, then check your mix on speakers. Uploading an unlisted video to YouTube and playing it through a TV soundbar is now, as he calls it, a very good “car listen” for immersive mixes.

6. One Mix Is Rarely Enough – So Carry Several

Folding an immersive mix down to stereo always involves compromise. Past 5.1, Klaus often has to choose between making the fold-down sound its best or making the immersive mix as exciting as possible. Eclipsa lets you deliver a dedicated stereo mix and a dedicated immersive mix in one stream, which he sees as a huge advantage.

For film and TV, Scott explains, the same feature supports alternate languages, commentary and bonus content in one file. Most productions can only afford a single mix, so mixers rely on automated fold-downs, sometimes with their own custom recipe. At Netflix, he set up cloud transcoding that turned one master file into Atmos, spatial audio and stereo versions, and Eclipsa could work the same way.

7. Dialogue You Can Actually Hear

Josh admits to turning on subtitles just to follow the dialogue, and Scott says he’s far from alone. Dialogue intelligibility can go wrong anywhere between the set and the living room, from subtle modern acting styles to tiny TV speakers. An AES group he’s part of has published a paper covering the whole chain.

Next-generation formats like Eclipsa open up a real fix. Dialogue, music and effects can be streamed as separate elements and mixed on the device, which enables dialogue boost settings, a choice of sports commentators or none at all, and even a full personal mix for enthusiasts.

8. What Device Makers and Developers Need

For hardware makers, supporting Eclipsa comes down to software: a decoder plus a renderer for speakers or headphones. Samsung supports it on its 2025 TVs and newer soundbars, LG has started shipping it, Android 16 includes the decoder, and a Chrome prototype was already running at the time of recording. HDMI pass-through standardisation is in progress so soundbars and receivers can receive the stream directly.

For developers, the entry point depends on what you’re building. Jani recommends the Alliance for Open Media GitHub, particularly the iamf-tools repository, for the core technology, and Android’s ExoPlayer for building a player app. Anyone can contribute to the open source code, and companies and universities can join AOMedia to help shape future versions of the specification.

9. Live Is Coming

There’s no live Eclipsa streaming service yet, but the open source encoder can already encode in real time. Samsung has been testing live content with K-pop acts, the University of York is researching live applications, and Klaus is working on immersive stems for DJs and live wellness and meditation broadcasts.

His hope is for an immersive Boiler Room experience, where remote listeners feel part of the crowd.

Closing

Josh closes by asking each guest what success would look like. For Jani, it’s a format that plays on almost any device, filled with great content, because content is king. For Scott, it’s mixes reaching more people as the creators intended, not in a compromised version. For Klaus, it’s the day someone outside the industry tells him about an amazing immersive mix they heard on YouTube.

What ties the conversation together is the belief that opening up the format will do what licensing never could: put immersive audio in the hands of millions of creators and developers, and see what they build with it.

‍

‍

‍