A library for using Ogg Vorbis in web browsers
npm install sl-web-oggA library for using Ogg Vorbis in web browsers with the following features:
* Decodes and encodes Ogg Vorbis files, including comments.
* Decoding Ogg Vorbis audio data with browser-native implementation. (It's just faster.)
* No file-hosting requirements to use API. Just import the package and use it.
* Asynchronous promise-based API rather than use of web workers. (That may or may not be good for what you need.)
* You don't need to install a C toolchain to build the C-code portion of the project. Just use the pre-built WASM.
* Zero dependencies other than Xiph reference implementations in C via .gitmodules.
* Compatible with Webpack 5+ builds. (In other words, it does not rely on those deprecated Node-style definitions (e.g. Process) from Webpack 4 that are often declared as external dependencies causing breakage.)
Why would I make another Ogg package? Every other package I looked at was lacking one or more of the features above. And maybe I'm biased towards writing my own code. Well, I definitely am biased towards writing my own code.
https://github.com/erikh2000/sl-web-ogg
The examples below give a quick idea of how to accomplish different things. For a full working example of encoding, see the simple web app under the /example directory.
javascript
import { encodeAudioBuffer } from 'sl-ogg'; //...load an audio file into the audioBuffer variable.
encodeAudioBuffer(audioBuffer).then(oggBlob => {
// You've got a Blob. It can be downloaded, saved to persistent storage, or uploaded to a server - whatever you want to do with it.
});
`$3
javascript
import { encodeAudioBuffer } from 'sl-ogg'; //...load an audio file into the audioBuffer variable.
const commentTags = [{tag:'genre', value:'alternative'}, {tag:'artist', value:'REALLIFEALWAYS'}];
encodeAudioBuffer(audioBuffer, {quality:1, commentTags}).then(oggBlob => {
//...do your stuff with blob.
});
`$3
Browser-native functionality is used to decode the Ogg Vorbis file because it's almost always going to be faster than even WASM code. With this approach, a user gesture will be needed on some browsers before or decodeOggBlobWithTags() can be successfully called. In other words, the user has to click on something (anything) before the code below will work.`javascript
import { decodeOggBlob } from 'sl-ogg'; //...load an Ogg file into blob variable.
decodeOggBlob(blob).then(audioBuffer => {
// You've got an AudioBuffer. It can be played, samples analyzed, etc.
});
`$3
In some browsers (frigging Safari is the Internet Explorer of 202x!), there is a limit on the number of AudioContext instances you can have open. To avoid hitting this limit unexpectedly, you can hold on to one instance and pass it to so that only this single instance is used.
`javascript
import { decodeOggBlob } from 'sl-ogg'; //...load an Ogg file into blob variable.
const audioContext = new AudioContext();
decodeOggBlob(blob, audioContext).then(audioBuffer => {
//...do your stuff with audioBuffer
});
`$3
You can put comments into an Ogg file that basically can be any text data you want.
javascript
import { decodeOggBlob } from 'sl-ogg'; //...load an Ogg file into blob variable.
decodeOggBlobWithTags(blob).then(([audioBuffer, tags]) => {
console.log(tags); // It's an array of {name:string, value:string} elements.
});
`I decided against adding a documentation generator to this project. (It's an unneeded dependency.) But you can look at JSDoc-style comments in
to see the full API documentation.Building
If you want to build the JS portion of the library, rather than just importing it from NPM, do this:
1. Clone this repo.
2.
npm install
3. /dist will contain the files you'd need to bundle/redistribute with your app.
If you want to run the example app after following the previous steps to build:
1.
If you are paranoid like me, and want to build your own WASM: (What crazy malware might this Erik guy have inserted into that black box?!?)
1.
I realize step 2 is kind of cheeky. But I honestly can't remember the different things I had to install along the way, and to recreate the clean environment to figure it out is hard. The main thing you are installing is Emscripten, so the instructions here may be helpful.
I'm quite willing to update these instructions to make them better. Feel free to holler at me in Github issues if you see something about them that could be better or you are just stuck.
There's three layers to this thing.
API Layer* (JS) - the thing your web app calls. It wraps the Middle Layer's API, and unlike the Middle Layer has access to DOM, promise mechanics, and browser-provided web APIs.
Middle Layer* (C) - some functions that abstract over the fairly low-level Vorbis APIs to handle encoding and decoding. Unlike the API Layer, it has direct memory access and can concisely and performantly manipulate variables to call the Vorbis APIs.
Vorbis Layer* - the WASM-compiled libvorbis and libogg. These sources are used unchanged from pinned versions of Xiph reference implementations via Git submodules. (.gitmodules)
* Low dependencies, preferably none.
* Any dependencies added are evaluated for security and maintenance issues.
* Functions and variables are named for clarity and consistency.
* Explain the reasons for doing things in comments even if it means revealing mistakes. In other words, invite understanding and disagreement.
To honor the licenses of the Xiph reference implementations, I've copied the COPYING files from the Ogg and Vorbis repositories, respectively, to Ogg-COPYING and Vorbis-COPYING. This would cover the distribution of the WASM binaries under this repository. I think I am not technically distributing any source code from the Xiph reference implementations with my use of .gitmodules. But if I were, then including the COPYING files should satisfy the licensing terms.
All other files in the project are licensed under the MIT license.
middle-layer.c is a copy-and-rewrite from Yuji Miyane's code at github.com/higuma/ogg-vorbis-encoder-js/blob/master/src/encoder.c, which has its own prior contributors that Yuji also based his work on. Yuji's project is MIT-licensed like mine, so I believe my licensing is compatible with his wishes.
For anyone from Xiph or elsewhere that feels I've not followed licensing correctly, please file a Github issue or otherwise contact me. I aim to do right, and I am on the side of those who create.
The project isn't open to contributions at this point. But that could change. Contact me if you'd like to collaborate.
You can reach me on LinkedIn. I'll accept connections if you will just mention "sl-ogg" or some other shared interest in your connection request.
https://www.linkedin.com/in/erikhermansen/