Lossy or lossless: which audio format should I keep my library in?

For long-term archiving choose FLAC (lossless, ~50% the size of WAV). For everyday listening on phones and streaming use MP3 320 kbps or Opus 192 kbps (transparent quality, no audible difference at normal volumes). Avoid converting lossy → lossy → lossy: every step compounds artifacts.

Lossy or lossless: which audio format should I keep my library in?

For long-term archiving choose FLAC (lossless, ~50% the size of WAV). For everyday listening on phones and streaming use MP3 320 kbps or Opus 192 kbps (transparent quality, no audible difference at normal volumes). Avoid converting lossy → lossy → lossy: every step compounds artifacts.

Guide

AAC Audio Format: The Complete Technical Guide

PC By Pablo Cirre • Updated May 7, 2026

AAC Audio Format: The Complete Technical Guide

AAC (Advanced Audio Coding) is the dominant lossy audio codec for mainstream digital distribution, streaming, and mobile devices. Standardized as ISO/IEC 13818-7 (MPEG-2 AAC) in 1997 and extended in MPEG-4 (ISO/IEC 14496-3), AAC was designed from the ground up to surpass MP3's limitations. At equal perceived quality, AAC achieves roughly 30% smaller file sizes than MP3 — a 192 kbps AAC file sounds comparable to a 256 kbps MP3.

Historical Context and Standardization

AAC emerged from the MPEG Audio work that produced MP3 but incorporated significant advances from Dolby Laboratories, AT&T Bell Labs, Sony, Nokia, and Fraunhofer IIS. The original MPEG-2 AAC (1997) improved on MP3 in several ways: more filter bank channels, temporal noise shaping (TNS), perceptual noise substitution (PNS), and better joint stereo modes. MPEG-4 AAC (1999) extended it further with Scalable Sample Rate (SSR), Long Term Prediction (LTP), and eventually the HE-AAC (High-Efficiency AAC) family. Apple adopted AAC as the default format for iTunes in 2003, cementing its mainstream position. YouTube, Spotify, Apple Music, Netflix, and virtually all streaming services use AAC as their primary delivery codec.

AAC Profiles and Object Types

AAC is not a single codec but a family of profiles:

AAC-LC (Low Complexity): The most widely used profile. Requires modest CPU for decoding, universal hardware support. Used by iTunes, Apple Music, YouTube, most streaming services. The baseline for all AAC comparison.

HE-AAC v1 (High-Efficiency AAC, aacPlus v1): AAC-LC core + Spectral Band Replication (SBR). SBR encodes only the lower half of the spectrum in the AAC-LC core and reconstructs the upper half at decode time using spectral envelope information. Excellent at 48–80 kbps — sounds better than AAC-LC at 128 kbps at these bitrates. Used for DAB+ digital radio, internet radio streams, podcast delivery.

HE-AAC v2 (aacPlus v2): HE-AAC v1 + Parametric Stereo (PS). PS encodes a mono downmix plus stereo parameters (inter-channel level differences, inter-channel phase differences, inter-channel coherence). Usable down to 24–32 kbps — the lowest quality audio that still sounds like stereo. Used for lowest-bitrate internet radio streams.

AAC-LD (Low-Delay): Algorithmic delay reduced to ~20 ms (vs. ~60 ms for AAC-LC). Used for real-time communication. Used in FaceTime audio calls.

AAC-ELD (Enhanced Low-Delay): Combines AAC-LD with SBR. Delay as low as ~15 ms. Used in Apple FaceTime HD, professional IP audio.

xHE-AAC (Extended HE-AAC / USAC): MPEG-D Unified Speech and Audio Coding (ISO/IEC 23003-3). Combines AAC-ELD, HE-AAC v2, and new tools (ACELP coding for speech, harmonic transient shaping). Quality comparable to Opus. Used in MPEG-H Audio for broadcasting.

Technical Architecture: AAC-LC

Filter Bank: Modified Discrete Cosine Transform

AAC-LC uses a pure MDCT filter bank (unlike MP3's hybrid polyphase+MDCT):

Long windows (N=2048 samples → 1024 spectral lines): Normal mode, high frequency resolution
Short windows (N=256 samples → 128 spectral lines, 8 per long-window period): Transient mode, high time resolution
Start/Stop transition windows: Smooth switching between long and short blocks

The larger MDCT size (1024 vs. MP3's 576 lines) provides better frequency resolution and more efficient coding of stationary audio.

Psychoacoustic Model and Bit Allocation

AAC uses a more sophisticated psychoacoustic model than MP3:

1. FFT analysis: 2048-point FFT for masking threshold computation, simultaneous with MDCT 2. Spreading function: Models upward and downward masking across critical bands (Bark scale, 49 short-term masking bands) 3. Temporal masking: Pre-masking (2 ms before masker) and post-masking (up to 200 ms after masker) 4. Tonality estimation: Distinguishes tonal components (sinusoids) from noise components for different masking thresholds 5. Rate-distortion optimization: Allocates bits to minimize perceptual distortion subject to bit budget

Temporal Noise Shaping (TNS)

TNS is one of AAC's key advances over MP3. It applies LPC prediction in the frequency domain to shape the temporal distribution of quantization noise. For signals with strong tonal components that vary rapidly over time (speech, drums, piano attacks), TNS prevents the psychoacoustic model from producing pre-echo artifacts by concentrating noise within the signal's temporal masking region.

Perceptual Noise Substitution (PNS)

For spectral bands deemed to be noise-like (non-tonal), AAC replaces the actual spectral coefficients with a synthesized noise signal of matching power level. Only the noise energy is transmitted — not the actual spectral shape. This frees bits for more complex tonal components, improving overall quality at low bitrates.

Joint Stereo: Mid/Side and Intensity Coding

Mid/Side (MS) stereo: Encodes L+R and L-R signals. Applied band-by-band based on which representation is more efficient. More flexible than MP3's whole-frame MS.
Intensity stereo: For high-frequency bands where the ear cannot localize, one channel's phase is discarded and only a single channel's magnitude is transmitted with a steering angle. Significant bit savings for content where stereo imaging above 6 kHz is imperceptible.

Scale Factors and Huffman Coding

AAC uses 49 scale factor bands (groups of spectral lines matched to critical bands). Scale factors are DPCM-coded relative to the previous band. Spectral coefficients within each scale factor band are quantized and coded using one of 11 unsigned Huffman codebooks (cb=1–10 for pairs of coefficients, cb=11 for quadruples of ±1,0 coefficients).

Bitrate Reference: AAC-LC

Bitrate	Perceived Quality	Equivalent MP3	Use Case
32 kbps	AM radio	48 kbps MP3	Podcast mono
64 kbps	Good stereo	96 kbps MP3	Streaming (podcast)
96 kbps	Transparent for most	128 kbps MP3	Standard streaming
128 kbps	High quality	192 kbps MP3	Apple Music 256 kbps (was 128)
192 kbps	Near-transparent	256 kbps MP3	High-quality streaming
256 kbps	Transparent	320 kbps MP3	Apple Music current standard
320 kbps	Archival lossy	—	Maximum AAC-LC

Apple Music streams at 256 kbps AAC-LC (encoded with Apple's proprietary AAC encoder, which is considered higher quality than open-source alternatives at the same bitrate).

Container Formats

AAC audio is stored in multiple container formats:

M4A (.m4a): MPEG-4 audio container (ISO Base Media File Format). The standard iTunes/Apple container. Contains one AAC audio track + metadata. Supports ID3v2-compatible iTunes metadata (iTunNORM, thumbnail, etc.).
MP4 (.mp4): Same as M4A but typically contains video + audio. AAC is the standard audio track in MP4 video.
ADTS (.aac): Audio Data Transport Stream. Raw AAC frames with a 7–9 byte sync header per frame. Streamable but no container metadata. Simpler than M4A, used in HLS segment delivery.
ADIF (.aac): Audio Data Interchange Format. Single header block at file start. For file delivery, not streaming.
Ogg (.oga): AAC in Ogg container. Uncommon but valid.

Encoding Commands

# FFmpeg: AAC-LC at 192 kbps (using native encoder)
ffmpeg -i input.wav -c:a aac -b:a 192k output.m4a

# FFmpeg: AAC using libfdk_aac (higher quality encoder, requires FFmpeg rebuild)
ffmpeg -i input.wav -c:a libfdk_aac -b:a 256k output.m4a

# FFmpeg: HE-AAC v1 at 64 kbps (libfdk_aac required)
ffmpeg -i input.wav -c:a libfdk_aac -profile:a aac_he -b:a 64k output.m4a

# FFmpeg: HE-AAC v2 at 32 kbps
ffmpeg -i input.wav -c:a libfdk_aac -profile:a aac_he_v2 -b:a 32k output.m4a

# FLAC to AAC-LC 256 kbps (Apple-quality archival)
ffmpeg -i input.flac -c:a aac -b:a 256k output.m4a

# Extract AAC from MP4 without transcoding
ffmpeg -i video.mp4 -c:a copy -vn output.m4a

# Batch convert WAV to AAC 192 kbps
for f in *.wav; do ffmpeg -i "$f" -c:a aac -b:a 192k "${f%.wav}.m4a"; done

AAC vs MP3 vs Opus: When to Use Each

Scenario	Best Format	Reason
Legacy device compatibility	MP3	Universal hardware support
Apple ecosystem (iOS, iTunes)	AAC M4A	Native support, best encoder
Web streaming (modern browsers)	Opus	Better quality at 96–128 kbps
Podcast distribution	AAC-LC 128 kbps	iTunes/Spotify compatibility
Low-bitrate internet radio	HE-AAC v2 32 kbps	Best quality at ultra-low bitrates
Long-term archival	FLAC	Lossless, future-proof
WebRTC/VoIP	Opus	Designed for real-time, low latency

Related conversions

Audio format pairs that come up most often:

Related conversions

Put what you just learned into practice — convert your files now in seconds, free and without registration.

Convert →

Convert →

Convert →

Convert →

Convert →

Convert →

Frequently Asked Questions

Yes, AAC-LC generally produces perceptually better quality than MP3 at the same bitrate. AAC uses a more sophisticated psychoacoustic model, a larger MDCT transform (1024 vs 576 spectral lines), temporal noise shaping (TNS), and more flexible joint stereo. A 128 kbps AAC file is commonly considered equivalent to a 192 kbps MP3 in perceived quality. Apple Music uses 256 kbps AAC-LC as their high-quality tier.

Sim, AAC-LC Geralmente produces perceptually better quality than MP3 at the same bitrate. AAC uses a more sophisticated psychoacoustic model, a larger MDCT transform (1024 vs 576 spectral lines), temporal noise shaping (TNS), e more flexible joint stereo. A 128 kbps AAC arquivo é commonly considered equivalent para um 192 kbps MP3 in perceived quality. Apple Music uses 256 kbps AAC-LC as their alta-quality tier.

Ja, AAC-LC Generell produces perceptually better quality than MP3 at the same Bitrate. AAC uses a more sophisticated psychoacoustic model, a larger MDCT transform (1024 vs 576 spectral lines), temporal noise shaping (TNS), und more flexible joint stereo. A 128 kbps AAC Datei is commonly considered equivalent zu einem 192 kbps MP3 in perceived quality. Apple Music uses 256 kbps AAC-LC as their hoch-quality tier.

Sí, AAC-LC Generalmente produces perceptually better quality than MP3 at the same bitrate. AAC uses a more sophisticated psychoacoustic model, a larger MDCT transform (1024 vs 576 spectral lines), temporal noise shaping (TNS), y more flexible joint stereo. A 128 kbps AAC archivo is commonly considered equivalent a un 192 kbps MP3 in perceived quality. Apple Music uses 256 kbps AAC-LC as their alta-quality tier.

For long-term archiving choose FLAC (lossless, ~50% the size of WAV). For everyday listening on phones and streaming use MP3 320 kbps or Opus 192 kbps (transparent quality, no audible difference at normal volumes). Avoid converting lossy → lossy → lossy: every step compounds artifacts.

AAC-LC (Low Complexity) is the standard profile used by iTunes and most streaming services — good quality above 96 kbps. HE-AAC v1 adds Spectral Band Replication (SBR), which reconstructs high frequencies at decode time, making it excellent at 48–80 kbps (digital radio quality). HE-AAC v2 adds Parametric Stereo on top, enabling usable stereo down to 24–32 kbps for internet radio streams. As bitrate increases above 96 kbps, AAC-LC becomes more efficient than HE-AAC variants.

AAC-LC (Low complexoity) is the padrão profile used by iTunes e most streaming services — boa qualidade above 96 kbps. HE-AAC v1 adds Spectral Band Replication (SBR), which reconstructs alta frequencies at decode time, making it excellent at 48–80 kbps (digital radio quality). HE-AAC v2 adds Parametric Stereo on top, enabling usable stereo down to 24–32 kbps para internet radio streams. As bitrate increases above 96 kbps, AAC-LC becomes more efficient than HE-AAC variants.

AAC-LC (Low Complexity) is the Standard profile used by iTunes und most streaming services — gut quality above 96 kbps. HE-AAC v1 adds Spectral Band Replication (SBR), which reconstructs hoch frequencies at decode time, making it excellent at 48–80 kbps (digital radio quality). HE-AAC v2 adds Parametric Stereo on top, enabling usable stereo down to 24–32 kbps für internet radio streams. As Bitrate increases above 96 kbps, AAC-LC becomes more efficient than HE-AAC variants.

AAC-LC (Low Complexity) is the estándar profile used by iTunes y most streaming services — bueno quality above 96 kbps. HE-AAC v1 adds Spectral Band Replication (SBR), which reconstructs alta frequencies at decode time, making it excellent at 48–80 kbps (digital radio quality). HE-AAC v2 adds Parametric Stereo on top, enabling usable stereo down to 24–32 kbps para internet radio streams. As bitrate increases above 96 kbps, AAC-LC becomes more efficient than HE-AAC variants.

Spoken-word podcasts: 44.1 kHz mono, 64–96 kbps Opus or 96 kbps MP3. Music podcasts: 44.1 kHz stereo, 128–192 kbps. Going above wastes bandwidth — speech has limited frequency content above 7 kHz, and listeners on data plans appreciate the smaller file. Apple Podcasts and Spotify both accept up to 48 kHz / 320 kbps.

Apple's proprietary AAC encoder (used in iTunes, Logic Pro, and Core Audio) is widely considered the best at low-to-medium bitrates. Among open-source options, libfdk_aac (Fraunhofer FDK AAC) outperforms FFmpeg's native AAC encoder. FFmpeg must be compiled with --enable-libfdk-aac to use it. For quick encoding without quality-critical requirements, FFmpeg's native encoder is adequate above 192 kbps.

Apple's proprietary AAC encoder (used in iTunes, Logic Pro, e Core Audio) is widely considered the best at baixa-to-medium bitrates. Among open-source options, libfdk_aac (Fraunhofer FDK AAC) outperforms FFmpeg's native AAC encoder. FFmpeg must be compiled com --enable-libfdk-aac to usar it. para quick codificação sem quality-critical requirements, FFmpeg's native encoder is adequate above 192 kbps.

Apple's proprietary AAC encoder (used in iTunes, Logic Pro, und Core Audio) is widely considered the best at niedrig-to-medium Bitrates. Among open-source options, libfdk_aac (Fraunhofer FDK AAC) outperforms FFmpeg's native AAC encoder. FFmpeg must be compiled mit --enable-libfdk-aac to verwenden it. für quick Codierung ohne quality-critical requirements, FFmpeg's native encoder is adequate above 192 kbps.

Apple's proprietary AAC encoder (used in iTunes, Logic Pro, y Core Audio) is widely considered the best at baja-to-medium bitrates. Among open-source options, libfdk_aac (Fraunhofer FDK AAC) outperforms FFmpeg's native AAC encoder. FFmpeg must be compiled con --enable-libfdk-aac to usar it. para quick codificación sin quality-critical requirements, FFmpeg's native encoder is adequate above 192 kbps.

Lossy → lossy compounds quantization noise. Each encode discards the same kind of perceptual information again, multiplying artifacts. Always re-encode from a lossless master if you have one (WAV, FLAC, or the original recording). If only an MP3 is available, keep the bitrate at or above the source — never go up to "improve quality".

.m4a is an MPEG-4 audio container (ISO Base Media File Format) containing AAC audio streams plus full metadata support (title, artist, album art, chapters). .aac files typically use ADTS (Audio Data Transport Stream) format — raw AAC frames with minimal headers, designed for streaming but lacking rich metadata support. For music files, .m4a is preferred. For HLS streaming segments, .aac ADTS is common. Both contain AAC audio data.

.m4a is an MPEG-4 audio container (ISO Base Media formatoo de arquivo) containing AAC audio streams plus full metadata support (title, artist, album art, chapters). .aac arquivos tipicamente usar ADTS (Audio Data Transport Stream) formato — raw AAC frames com minimal headers, designed para streaming mas lacking rich metadata support. para music files, .m4a is preferred. para HLS streaming segments, .aac ADTS is common. Both contain AAC audio data.

.m4a is an MPEG-4 audio Container (ISO Base Media Dateiformat) containing AAC audio streams plus full metadata support (title, artist, album art, chapters). .aac Dateien typically verwenden ADTS (Audio Data Transport Stream) Format — raw AAC frames mit minimal headers, designed für streaming aber lacking rich metadata support. für music files, .m4a is preferred. für HLS streaming segments, .aac ADTS is common. Both contain AAC audio data.

.m4a is an MPEG-4 audio contenedor (ISO Base Media formatoo de archivo) containing AAC audio streams plus full metadata support (title, artist, album art, chapters). .aac archivos typically usar ADTS (Audio Data Transport Stream) formato — raw AAC frames con minimal headers, designed para streaming pero lacking rich metadata support. para music files, .m4a is preferred. para HLS streaming segments, .aac ADTS is common. Both contain AAC audio data.

Most modern tools (FFmpeg with <code>-map_metadata 0</code>, foobar2000, dBpoweramp) preserve ID3 tags and embedded cover art. Some quick-and-dirty converters strip them silently. KaijuConverter preserves tags in its <a href="/convert/flac-to-mp3">audio conversions</a>; if metadata is critical to you, verify the output with <code>ffprobe</code> before deleting the original.