AAC (Advanced Audio Coding) — Complete Format Guide
What Is AAC?
AAC (Advanced Audio Coding) is a lossy audio compression standard developed as the successor to MP3 by Fraunhofer IIS, AT&T Bell Laboratories, Dolby Laboratories, Sony, and Nokia. Standardised in 1997 as MPEG-4 Audio (ISO/IEC 13818-7), AAC became the mandatory audio format for the MPEG-4 video standard and has since become the dominant audio codec in mobile devices, streaming platforms, and consumer electronics.
Apple selected AAC as the iTunes Store's audio format in 2003, cementing its mainstream adoption. Today AAC is the default audio codec for:
- Apple ecosystem — iTunes, Apple Music, iPhone/iPad video, FaceTime, QuickTime
- YouTube — AAC audio track in MP4 containers
- PlayStation and Xbox — game audio and multimedia
- Digital broadcasting — DAB+, DVB, ATSC 2.0
- Streaming — Amazon Music, SoundCloud, most video conferencing platforms
AAC Profiles
Unlike MP3 (a single codec), AAC is a family of profiles optimised for different use cases:
AAC-LC (Low Complexity) — The Standard
The baseline AAC profile. Balanced quality-vs-complexity trade-off:
- Target bitrate: 96–256 kbps for stereo music
- Complexity: moderate; hardware decoders available in all modern chips
- Quality: superior to MP3 at equivalent bitrates; transparent at 192–256 kbps
- Use case: music streaming, podcast distribution, stored audio
HE-AAC v1 (High Efficiency) — Spectral Band Replication
Adds Spectral Band Replication (SBR) on top of AAC-LC:
- SBR encodes only the low-frequency half of the spectrum (~11 kHz) at full resolution; high frequencies are reconstructed at the decoder from the low-frequency data using parametric synthesis
- Target bitrate: 32–96 kbps for stereo
- Quality at low bitrates dramatically better than AAC-LC; audio sounds "fuller" at 48 kbps vs. AAC-LC at same bitrate
- Use case: internet radio, mobile streaming, broadcast radio (DAB+)
HE-AAC v2 — Parametric Stereo
Adds Parametric Stereo (PS) to HE-AAC v1:
- PS encodes the audio as mono, then adds stereo reconstruction parameters (inter-channel level difference, inter-channel phase difference, inter-channel coherence) that the decoder uses to reconstruct the stereo image
- Target bitrate: 16–64 kbps for stereo (remarkable quality at these bitrates)
- Use case: satellite radio, internet radio at very low bitrates, in-flight entertainment
xHE-AAC (Extended High Efficiency) — USAC
The most modern AAC profile (2012), also known as USAC (Unified Speech and Audio Coding):
- Combines AAC-LC, SBR, PS, and a new speech-optimised coding mode (ACELP) that adapts between music and speech automatically
- Target bitrate: 12–64 kbps for stereo
- Quality at ultra-low bitrates (16–32 kbps) dramatically superior to all previous audio codecs
- Use case: mobile streaming at constrained bandwidth, digital broadcast replacement
How AAC Compresses Audio
AAC builds on the psychoacoustic principles of MP3 but with significant improvements:
Filterbank: Modified Discrete Cosine Transform (MDCT)
AAC uses MDCT with window sizes of 1024 samples (long block) or 128 samples (short block, used during transients). The filterbank provides 49 frequency bands vs. MP3's 576 subbands — more fine-grained spectral analysis.
Temporal Noise Shaping (TNS)
TNS applies a prediction filter across frequency coefficients to shape quantisation noise so it follows the amplitude envelope of the audio signal. This dramatically reduces pre-echo artefacts (ringing before transient events like drum hits) that plague MP3 at low bitrates.
Perceptual Noise Substitution (PNS)
For noise-like audio regions (cymbals, breath sounds, high-frequency detail), rather than encoding actual spectral coefficients, AAC encodes only energy level and spectral shape, then reconstructs plausible noise at the decoder. This saves bits without perceptible quality difference.
Mid-Side Stereo and Intensity Stereo
AAC jointly codes stereo channels using M/S stereo (encodes sum/difference signals rather than left/right) and intensity stereo (at high frequencies where human spatial localisation is imprecise, encodes only energy rather than full spectral detail). MP3 supports M/S stereo but AAC implements it more flexibly at the band level.
Long-Term Prediction (LTP)
For periodic/tonal audio (held musical notes, vowels), LTP uses the periodicity from previous frames to predict current samples, dramatically reducing residual energy that must be explicitly coded.
Containers for AAC Audio
AAC bitstreams are packaged in containers:
| Container | Extension | Common Use |
|---|---|---|
| MPEG-4 Audio | .m4a | iTunes, Apple Music, podcast |
| MPEG-4 Video | .mp4 | YouTube, most video with AAC audio |
| ADTS | .aac | Raw AAC with sync headers; streaming |
| 3GPP / 3G2 | .3gp / .3g2 | Mobile video (legacy) |
| MPEG-TS | .ts | Broadcast, DVB, HLS streaming |
The .m4a extension is an Apple convention for MPEG-4 Audio — it distinguishes audio-only AAC from video-containing .mp4 files. Technically they use the same container format.
Quality and Bitrate Guide
AAC-LC for Music
| Bitrate | Quality | Use Case |
|---|---|---|
| 64 kbps | Acceptable | Background music, low-bandwidth streaming |
| 96 kbps | Good | Radio-quality podcast, voice content |
| 128 kbps | Very good | Most listeners satisfied for casual music |
| 192 kbps | Excellent | Audiophile casual; hard to distinguish from lossless |
| 256 kbps | Near-transparent | Apple Music and streaming service standard |
| 320 kbps | Transparent | Exceeds most listeners' detection threshold |
Apple Music streams at 256 kbps AAC-LC for standard quality. Apple Music Lossless streams ALAC (Apple Lossless) at the same price — demonstrating AAC-LC 256 kbps is Apple's recommended quality floor for music.
Advantages of AAC
- Better quality than MP3 — at any given bitrate, AAC typically delivers superior sound; particularly noticeable below 192 kbps
- Universal support — every smartphone, tablet, smart TV, streaming service, and car audio system
- Apple ecosystem native — zero compatibility issues on iOS, macOS, tvOS, watchOS
- Multiple profiles — scales from 12 kbps (xHE-AAC) to 320 kbps+ (AAC-LC) with appropriate profile selection
- Low-power hardware decoding — dedicated AAC decode hardware in virtually every mobile chip (saves battery vs. software decode)
- Surround sound — AAC supports 5.1, 7.1, and object-based surround audio
- Standardised — maintained by ISO; not controlled by a single vendor
Disadvantages of AAC
- Not completely open — AAC encoders and decoders require patent licences from MPEG LA; free/open-source decoders (libfaad) exist but encoder quality varies
- Complexity — multiple profiles with different feature sets; cross-profile playback compatibility not always guaranteed
- Not lossless — discards audio information permanently; not suitable for mastering or archival
AAC vs. MP3: Key Differences
| Feature | AAC | MP3 |
|---|---|---|
| Standard | MPEG-4 | MPEG-1/2 Layer 3 |
| Max channels | 48 | 2 (stereo) |
| Frequency range | Full bandwidth preserved | 16 kHz cutoff at low bitrates |
| Quality at 128 kbps | Excellent | Good |
| Quality at 96 kbps | Good | Acceptable with artefacts |
| TNS (pre-echo reduction) | Yes | No |
| PNS | Yes | No |
| Universal hardware support | Excellent | Excellent |
Converting AAC Files
AAC → MP3: Common for hardware players without AAC support (vintage devices, some car stereos); re-encoding from lossy to lossy causes generation loss; use highest quality AAC → FLAC: Lossless container for lossy content; retains the AAC artefacts but standardises storage; useful for library management AAC → WAV: Full PCM decode for audio editing; preserves AAC artefacts in PCM form MP3 → AAC: For re-encoding old MP3 libraries; note transcoding penalty; best to source from FLAC/WAV originals
Related conversions
Audio format pairs that come up most often: