The following code fails to compile on the latest commit of bevy_kira_audio (b721820). This issue is very similar to #31, however I can't find a commit that works for me. I absolutely need the feature introduced on this commit, but no commit after this one compiles successfully.

I also do not understand why this error is thrown in the first place. this code clearly shows that TypeUuid is defined for AudioSource so I'm not sure where the error is coming from.

Failing code:

• NiklasEi/bevy_game_template
• bevy and bevy_kira_audio set to the latest commit

#[derive(AssetCollection)]
pub struct AudioAssets {
    #[asset(path = "audio/flying.ogg")]
    pub flying: Handle<AudioSource>,
}

Error:

error[E0277]: the trait bound `AudioSource: TypeUuid` is not satisfied
   --> src/loading.rs:34:17
    |
34  |     pub flying: Handle<AudioSource>,
    |                 ^^^^^^^^^^^^^^^^^^^ the trait `TypeUuid` is not implemented for `AudioSource`
    |

Cargo.toml

...

[features]
default = [
    "bevy_kira_audio/ogg",
    "bevy/bevy_winit",
    "bevy/render",
    "bevy/png",
    "bevy/x11",
]

dev = ["bevy/dynamic"]

[dependencies]
bevy = { git = "https://github.com/bevyengine/bevy", rev = "7d554146524987b050923841de06c7d18adf66e9", default-features = false }
bevy_kira_audio = { git = "https://github.com/NiklasEi/bevy_kira_audio", rev = "b721820d0f7b479a136cc2a9d51c23bbedd2934e" }
bevy_asset_loader = "0.11.0"
rand = "0.8.5"

[target.'cfg(target_os = "linux")'.dependencies]
winit = { version = "0.26.1", features = ["x11"] }

[build-dependencies]
embed-resource = "1.7.2"

...

As discussed in #23

TODO

• [x] Get play state from Res<Audio>.state(handle), including: an enum with Playing, Paused, Stopped; a position as Option<f64>.
• [x] Merge InstanceHandlePriv into InstanceHandle
• [x] Rename InstanceState.priv_ => .handle
• [x] Write an example

I stress tested an application Im making (for fun really) and it plays audio. Due to the high volume of play commands added I started to get log messages that I think originates from the CommandQueueFull error (perhaps kira printed it?).

I did a profiling and ran the stress test and basically even long after all audio stops playing this method: https://github.com/NiklasEi/bevy_kira_audio/blob/a08ee2a4e3719cfbffbbca5f3feb61690e2233c6/src/audio_output.rs#L454 continues to take up in the profiling I did 34-35ms of frame-time (steadily growing from almost no time at start of app then really going haywire after a while). My guess is that the cleanup_stopped_instances don't detect them as faulty or even started but they still exist in the vector or something.

error[E0277]: the trait bound `AudioSource: TypeUuid` is not satisfied
  --> src/states/mod.rs:8:22
   |
8  |     pub loop_handle: Handle<AudioSource>,
   |                      ^^^^^^^^^^^^^^^^^^^ the trait `TypeUuid` is not implemented for `AudioSource`
   |
   = note: required because of the requirements on the impl of `Asset` for `AudioSource`
note: required by a bound in `bevy::prelude::Handle`
  --> /Users/visual/.cargo/git/checkouts/bevy-f7ffde730c324c74/91c3b21/crates/bevy_asset/src/handle.rs:65:8
   |
65 |     T: Asset,
   |        ^^^^^ required by this bound in `bevy::prelude::Handle`

For more information about this error, try `rustc --explain E0277`.

I saw the note in the readme about sound playback issues in Firefox and I just wanted to share my experiencewith my own Bevy Kira plugin on web ( I'm not using this plugin, but I am using Kira ).

I also had audio issues on Firefox, but I noticed that Firefox is also using almost 100% of one of my CPU cores. If I resize the browser window so that it is much smaller, though, the CPU usage goes down and the sound playback works nicely. That seems to indicate that the issue isn't actually with the sound playback itself, but how much CPU usage the Bevy game is using on Firefox. It seems that Bevy takes much more CPU on firefox than it does on Chrome.

I'm not sure exactly what causes the CPU usage difference, but at least I think that gives us somewhere to look. If resizing the screen changes the CPU usage, then it is probably rendering. Maybe firefox's rendering is just slower than chrome's.

I'm using a custom renderer, too, so my experience may not be the best benchmark, but that's what I know for what it's worth.

Hello again! Upgraded to latest version. Ended up being in a scenario where the amount of sounds played (not super-crazy) caused a panic here: https://github.com/NiklasEi/bevy_kira_audio/blob/main/src/audio_output.rs#L167

This is the full error message: thread 'main' panicked at 'Failed to play sound: SoundLimitReached'

I'm wondering if there is a way to get the limit by code? If so I guess I can implement a rate-limiter on sounds on my side. However I think it would be nice to propagate this error up to the user of bevy_kira_audio :)

Hello! 👋 Thanks for all your work! I really love this crate ❤️ ❤️

I'm creating this issue because I'd like to document some surprising (to me) behavior of bevy_kira_audio / kira in comparison to bevy_audio 😄

Because of some crackling audio in the browser (see #27) I recently ported a project of mine to bevy_audio in a branch to experiment.

After porting I noticed a significant speedup to load times as well as the crackling audio going away.

Anecdotally, when using bevy_kira_audio it takes a couple seconds for my "Play" button to display and then a couple more after clicking "Play" for the scene to render. When I switched to bevy_audio, everything much quicker, i.e. no delay between startup and "Play" button or after clicking "Play".

project code: https://github.com/shnewto/limbo_pass

wasm branches

• wasm-rodio
• wasm-kira

Some of this is definitely a candidate for being system specific, I'm on macOS Monterey 12.3.1. I'd be curious to know if this is or isn't the case for someone with a different setup.

Following bevyengine/bevy#2373 I would like to adapt licensing of this plugin to keep maximal compatibility with Bevy. For other reasons to switch to MIT/Apache 2.0 please take a look at the Bevy issue.

With only two other contributors this should be much easier than for Bevy itself :wink:

@rparrett and @CleanCut, I think you know the drill: please comment on this issue with

I license past and future contributions under the dual MIT/Apache-2.0 license, allowing licensees to choose either at their option.

Thanks again for your help!

Hey there!

Would you be open to a pull request that adds versions of the play, stop, etc. functions that additionally take the Tween parameter present in kira?

I was using my own wrapper around kira for Bevy, but would like to switch to this one and that looks like it's the only feature I'm missing.

I think we could re-export Tween from kira, but maybe rename it to AudioTween, just to avoid confusion with any other Bevy tweening plugins or something like that.

There are 3 slightly different cases. The common part is that we're playing a short sound on a separate AudioChannel with play() method.

• Waiting for the sound to end. Checking PlaybackState: it is permanently PlaybackState::Playing
• Waiting for the sound to end, running stop() method. Checking PlaybackState: it is permanently PlaybackState::Playing
• Not waiting for the sound to end, running stop() method while the sound is still playing. Checking PlaybackState: it is permanently PlaybackState::Stopping

In all cases my expectation for PlaybackState would be PlaybackState::Stopped.

Hi-

I'm trying to consume bevy directly from git, due to the fact the latest release 0.5.0 is 6 months old and is missing a few fixes I need (see bevyengine/bevy#2940). However I'm facing many build errors:

error[E0432]: unresolved import `bevy::prelude::AppBuilder`
error[E0432]: unresolved imports `bevy::core::AsBytes`, `bevy::render::pass::RenderPassDepthStencilAttachmentDescriptor`
error[E0261]: use of undeclared lifetime name `'w`
error[E0261]: use of undeclared lifetime name `'s`
error[E0107]: this struct takes 2 lifetime arguments but 1 lifetime argument was supplied

So I have a few questions:

• Is bevy_kira_audio only supposed to work with bevy on released version?
• Would it be possible to add some CI to catch those build issues along the way?

Thanks!

Hi,

Does kira_audio support having a single audio asset, but then having Audio handles that will read only parts of the audio file? (e.g. from 0.8sec to 1.5sec)

This would be helpful for the web as I would like to limit the number of files that clients have to download, so I'd like to 'merge' all audio files into one.

It is common to want to do something simple like stop the playback of a channel on entering or exiting a certain state. This PR removes boilerplate from this functionality. E.g. (using iyes_loopless):

use bevy::prelude::*;
use bevy_kira_audio::prelude::*;
use iyes_loopless::prelude::*;

#[derive(Debug, Clone, Copy, PartialEq, Eq, Hash)]
enum GameState {
    MainMenu
}

#[derive(Resource)]
struct BgmChannel;

struct FooPlugin;

fn main() {
    App::new()
        .add_loopless_state(GameState::MainMenu)
        .add_plugin(AudioPlugin)
        .add_audio_channel::<BgmChannel>()
        .add_plugin(FooPlugin)
        .run();
}

Current method:

impl Plugin for FooPlugin {
    fn build(&self, app: &mut App) {
        app.add_exit_system(GameState::MainMenu, stop_title_music);
    }
}

fn stop_title_music(audio: Res<AudioChannel<BgmChannel>>) {
    audio.stop();
}

With PR changes:

impl Plugin for FooPlugin {
    fn build(&self, app: &mut App) {
        app.add_exit_system(GameState::MainMenu, stop_playback::<BgmChannel>);
    }
}

Maybe I'm mistaken, but as far as I understand, it looks like kira supports lots of different tweens (allowing frequency slides etc.), but that the only exposed tween is through the fade_in command.

Or maybe exposing such features is out of scope?

Currently, to create a StaticSoundData for AudioSource, you have to import kira with the correct version. I'd like... to not do that lol.

Also, in the case that #63 would be implemented, it is likely that users would have to reimport kira just to implement their own SoundData.

A solution to this is to reexport kira itself.

Not sure to what extent this is a good idea, but I'm trying to manage various characteristics of dynamic channels. One missing feature I've noticed is the ability to iterate through all channels.

Might be useful to impl Iterator/IntoIter on DynamicChannels. Also, generally making it more hashmap-like would help as well (I.e. being able to retrieve keys/values.)

Thanks.