5 ms·

HRTFs are awesome, but I really wish we could have an open standard here and not a dozen competing proprietary ones. It feels like you really just need a standa

HRTFs are awesome, but I really wish we could have an open standard here and not a dozen competing proprietary ones. It feels like you really just need a standard way to encode the position of multiple audio tracks in 3D space and let your implementation of choice do the spatialization for you. Anyway, fun fact: OpenAL Soft has an implementation of this. If any games you play use it (Minecraft notably does) you can get 3D audio by just setting 'hrtf = true' in your OpenAL config.

colanderman 6y ago
> a standard way to encode the position of multiple audio tracks in 3D space Isn't that the idea behind Ambisonics [1], which OpenAL Soft is based on? [1] https://en.wikipedia.org/wiki/Ambisonics https://en.wikipedia.org/wiki/Ambisonics
- nitrogen 6y ago
  Ambisonics is for encoding a full-sphere soundfield. The components have a fixed location. Positioning individual tracks is not quite the same thing, as they can start/stop/move, and they are not limited to spherical harmonics. Video games have done exactly this for ages, but the only storage and distribution formats for that type of positional audio are still proprietary (e.g. Atmos).
  - colanderman 6y ago
    They should correct me if I'm wrong, but I believe the GP's point was that there should be a standardized way for software to present spatial audio to the sound system, which can then render it according to the user's preferences. How the sound is generated (e.g. moving point sources; spatial reverb; pre-spatialized tracks) does not matter. Yes, using even third-order Ambisonics is not as precise as rendering all point sources (and non-point sources...) exactly through the HRTF. But it is an existing standard you can use today via OpenAL Soft.
    - nitrogen 6y ago
      I'm currently working on something vaguely related, but presently focused on converting stereo to arbitrary N-channel surround (and documenting it on YouTube). I agree that spatialization should be more standardized, and it should be easy to convert between arbitrary spatialization setups.
anastasiadevana 6y ago
There is, except that every big company wants its own piece of the pie: https://adm.ebu.io/ https://adm.ebu.io/

colanderman 6y ago

> a standard way to encode the position of multiple audio tracks in 3D space Isn't that the idea behind Ambisonics [1], which OpenAL Soft is based on? [1] https://en.wikipedia.org/wiki/Ambisonics https://en.wikipedia.org/wiki/Ambisonics

nitrogen 6y ago
Ambisonics is for encoding a full-sphere soundfield. The components have a fixed location. Positioning individual tracks is not quite the same thing, as they can start/stop/move, and they are not limited to spherical harmonics. Video games have done exactly this for ages, but the only storage and distribution formats for that type of positional audio are still proprietary (e.g. Atmos).
- colanderman 6y ago
  They should correct me if I'm wrong, but I believe the GP's point was that there should be a standardized way for software to present spatial audio to the sound system, which can then render it according to the user's preferences. How the sound is generated (e.g. moving point sources; spatial reverb; pre-spatialized tracks) does not matter. Yes, using even third-order Ambisonics is not as precise as rendering all point sources (and non-point sources...) exactly through the HRTF. But it is an existing standard you can use today via OpenAL Soft.
  - nitrogen 6y ago
    I'm currently working on something vaguely related, but presently focused on converting stereo to arbitrary N-channel surround (and documenting it on YouTube). I agree that spatialization should be more standardized, and it should be easy to convert between arbitrary spatialization setups.

anastasiadevana 6y ago

There is, except that every big company wants its own piece of the pie: https://adm.ebu.io/ https://adm.ebu.io/