Most of the recent chainDRiVE work has gone into something quiet but important: the codec drawer. We wanted the list of formats it can actually encode to be both wider and honest — no fake “Supported” badges, no vague “High / Medium / Low.” Here is where that landed, and where we want to take the effects chain next.

The codec bench, refined

chainDRiVE now treats each format family as its own tool, with the parameters that actually matter surfaced where you can reach them:

  • AAC family — AAC-LC through native WebCodecs (with an fdk-aac WASM fallback), mono or stereo; plus HE-AAC v1 (SBR) and HE-AAC v2 (SBR + PS) via fdk-aac WASM, stereo-only. Written as ADTS .aac or MPEG-4 .m4a.
  • MP3 via LAME — CBR/VBR, an HQ toggle, six psychoacoustic/stereo profiles (LAME Default, FHG Narrow, Music, Speech, Wide Stereo, MS Joint Only), adjustable stereo width, and Force-MS joint stereo.
  • Lossless — FLAC, AIFF, AU/SND, and WAV PCM in 16- or 24-bit integer and 32-bit float.
  • Opus, Ogg & WebM — Opus (.opus) with Audio / VoIP / Low-Delay modes and complexity 0–10, plus Ogg Vorbis and WebM (Opus).
  • Voice & telephony — AMR-NB (8 kHz) and AMR-WB (16 kHz), WAV IMA & Microsoft ADPCM, G.711 A-law / μ-law, and GSM 6.10 — the low-bitrate and legacy codecs most converters skip.
  • Rates, channels & probing — sample rates from 8 kHz telephony up through 48 kHz, mono / stereo / joint stereo where the codec allows, size estimates before you commit, and capability probing so a format only appears when your device can genuinely encode it.

We also removed xHE-AAC. On paper it is attractive; in practice we could not encode it cleanly across the platforms we care about (fdk-aac cannot do USAC). Rather than dangle a broken option, we pulled it. If that changes, it comes back — but we would rather ship what actually works than pad a feature matrix.

What we want to add to the chain next

The formats are the foundation. The part we are most excited about is the effects chain — making it both easier to reach for and deeper when you want it. On the bench (intentions, not ship dates):

  • Compressor presets — one-tap starting points (voice, music, broadcast, and friends) instead of dialing six sliders from scratch every time.
  • A multiband compressor — independent dynamics across low / mid / high bands for more surgical control than a single-band pass.
  • Voice isolation — lift a voice up out of noise or backing music for cleaner speech.
  • EQ with presets — quick tonal shaping with saved curves you can drop onto a whole batch.
  • Maybe even an SBR-style restoration effect — spectral band replication that estimates and rebuilds the high frequencies missing from dull, low-bitrate, or low-fidelity recordings, so tired old audio sounds brighter and more open.

These are directions, not promises. Some will land as described, some will change shape, and the SBR restoration idea is firmly in the “let's see if we can pull it off” column. As always, we'll say what is real, what is partial, and what is still just on the bench.

Background reading: Codec Depth, chainDRiVE, and CHAiNAMP and Why Audio Transcoders Still Matter. Product surface while we build: chainDRiVE.