Symphonia v0.6, include all decoders/encoders always, refactors

.codespellignore

@@ -1,3 +1,4 @@
 # Exceptions, all lowercase, sorted lexicographically
+caf
 crate
 ser

CHANGELOG.md

@@ -1,5 +1,11 @@
 # Version History
 
+## Unreleased
+
+- Decoders are enabled always - the legacy cargo features mp3/ogg/flac/wav/.. are now no-op
+- Added .aiff decoder support
+- Fix: mp3 decoding would sometimes insert leading silent frames
+
 ## Version 1.4.0 (2026-05-18)
 
 - Added `BaseAudioContext::render_quantum_size` (128 always for now)

Cargo.toml

@@ -23,7 +23,11 @@ arc-swap = "1.6"
 arrayvec = "0.7"
 fft-convolver = "0.3"
 cpal = { version = "0.17", optional = true }
-creek = "1.2"
+creek = { version = "1.2", default-features = false, features = [
+    "decode",
+    "decode-all",
+    "encode-wav",
+] }
 crossbeam-channel = "0.5"
 cubeb = { version = "0.15", optional = true }
 dasp_sample = "0.11"
@@ -41,7 +45,10 @@ num-complex = "0.4"
 realfft = "3.3"
 rubato = "0.16"
 smallvec = "1.11"
-symphonia = { version = "0.5", default-features = false }
+symphonia = { version = "0.6", default-features = false, features = [
+    "all",
+    "opt-simd",
+] }
 vecmath = "1.0"
 
 [target.'cfg(any(target_arch = "x86", target_arch = "x86_64", target_arch = "aarch64"))'.dependencies]
@@ -69,24 +76,19 @@ path = "benches/my_benchmark.rs"
 harness = false
 
 [features]
-default = ["mp3", "ogg", "flac", "wav", "m4a", "alac", "cpal"]
-mp3 = ["symphonia/mp3", "creek/decode-mp3"]
-ogg = [
-    "symphonia/ogg",
-    "symphonia/vorbis",
-    "creek/decode-ogg",
-    "creek/decode-vorbis",
-]
-flac = ["symphonia/flac", "creek/decode-flac"]
-wav = ["symphonia/wav", "symphonia/pcm", "creek/decode-wav", "creek/decode-pcm"]
-aac = ["symphonia/aac", "creek/decode-aac"]
-m4a = ["aac", "symphonia/isomp4", "creek/decode-isomp4"]
-alac = [
-    "symphonia/alac",
-    "symphonia/isomp4",
-    "creek/decode-alac",
-    "creek/decode-isomp4",
-]
+default = ["cpal"]
+
+# Legacy codec feature names kept for backwards compatibility.
+# Decoding support is no longer split by file format; all supported Symphonia
+# and Creek decoders are enabled unconditionally.
+mp3 = []
+ogg = []
+flac = []
+wav = []
+aac = []
+m4a = []
+alac = []
+
 cpal = ["dep:cpal"]
 cubeb = ["dep:cubeb"]
 cpal-jack = ["cpal", "cpal/jack"]

README.md

@@ -117,6 +117,12 @@ via `cpal`'s `wasm-bindgen` backend. Check out [an example WASM
 project](https://github.com/orottier/wasm-web-audio-rs).
 Warning: experimental!
 
+## Audio decoding support
+
+Audio decoding is powered by Symphonia. The current implementation enables all
+Symphonia-supported audio formats and codecs, including AIFF, CAF, ISO/MP4,
+MKV/WebM, Ogg, WAV, AAC, ADPCM, ALAC, FLAC, MP1/MP2/MP3, PCM, and Vorbis.
+
 ## Contributing
 
 web-audio-api-rs welcomes contribution from everyone in the form of

src/context/base.rs

@@ -5,7 +5,7 @@ use crate::context::{
     AudioContextRegistration, AudioContextState, AudioParamId, ConcreteBaseAudioContext,
     DESTINATION_NODE_ID,
 };
-use crate::decoding::MediaDecoder;
+use crate::decoding::decode_media_data;
 use crate::events::{Event, EventHandler, EventType};
 use crate::node::{AudioNode, AudioNodeOptions};
 use crate::param::AudioParamDescriptor;
@@ -27,7 +27,9 @@ pub trait BaseAudioContext {
 
     /// Decode an [`AudioBuffer`] from a given input stream.
     ///
-    /// The current implementation can decode FLAC, Opus, PCM, Vorbis, and Wav.
+    /// The current implementation supports Symphonia's audio formats and codecs,
+    /// including AIFF, CAF, ISO/MP4, MKV/WebM, Ogg, WAV, AAC, ADPCM, ALAC,
+    /// FLAC, MP1/MP2/MP3, PCM, and Vorbis.
     ///
     /// In addition to the official spec, the input parameter can be any byte stream (not just an
     /// array). This means you can decode audio data from a file, network stream, or in memory
@@ -67,26 +69,14 @@ pub trait BaseAudioContext {
         &self,
         input: R,
     ) -> Result<AudioBuffer, Box<dyn std::error::Error + Send + Sync>> {
-        // Set up a media decoder, consume the stream in full and construct a single buffer out of it
-        let mut buffer = MediaDecoder::try_new(input)?
-            .collect::<Result<Vec<_>, _>>()?
-            .into_iter()
-            .reduce(|mut accum, item| {
-                accum.extend(&item);
-                accum
-            })
-            // if there are no samples decoded, return an empty buffer
-            .unwrap_or_else(|| AudioBuffer::from(vec![vec![]], self.sample_rate()));
-
-        // resample to desired rate (no-op if already matching)
-        buffer.resample(self.sample_rate());
-
-        Ok(buffer)
+        decode_media_data(input, self.sample_rate())
     }
 
     /// Decode an [`AudioBuffer`] from a given input stream.
     ///
-    /// The current implementation can decode FLAC, Opus, PCM, Vorbis, and Wav.
+    /// The current implementation supports Symphonia's audio formats and codecs,
+    /// including AIFF, CAF, ISO/MP4, MKV/WebM, Ogg, WAV, AAC, ADPCM, ALAC,
+    /// FLAC, MP1/MP2/MP3, PCM, and Vorbis.
     ///
     /// In addition to the official spec, the input parameter can be any byte stream (not just an
     /// array). This means you can decode audio data from a file, network stream, or in memory
@@ -109,23 +99,7 @@ pub trait BaseAudioContext {
            + Send
            + 'static {
         let sample_rate = self.sample_rate();
-        async move {
-            // Set up a media decoder, consume the stream in full and construct a single buffer out of it
-            let mut buffer = MediaDecoder::try_new(input)?
-                .collect::<Result<Vec<_>, _>>()?
-                .into_iter()
-                .reduce(|mut accum, item| {
-                    accum.extend(&item);
-                    accum
-                })
-                // if there are no samples decoded, return an empty buffer
-                .unwrap_or_else(|| AudioBuffer::from(vec![vec![]], sample_rate));
-
-            // resample to desired rate (no-op if already matching)
-            buffer.resample(sample_rate);
-
-            Ok(buffer)
-        }
+        async move { decode_media_data(input, sample_rate) }
     }
 
     /// Create an new "in-memory" `AudioBuffer` with the given number of channels,