Why is the audio out of sync after converting?

Common causes: (1) variable framerate source rendered as constant (use -vsync vfr to preserve VFR); (2) different audio sample rates not resampled (add -ar 48000 ); (3) container limitations (MP4 with variable framerate misbehaves — prefer MKV during editing, encode to MP4 only at the end). Always run ffprobe on both source and output to compare timing.

Why is the audio out of sync after converting?

Common causes: (1) variable framerate source rendered as constant (use -vsync vfr to preserve VFR); (2) different audio sample rates not resampled (add -ar 48000 ); (3) container limitations (MP4 with variable framerate misbehaves — prefer MKV during editing, encode to MP4 only at the end). Always run ffprobe on both source and output to compare timing.

Guide

FLAC: The Complete Guide to Free Lossless Audio Codec

PC By Pablo Cirre • Updated May 7, 2026

FLAC: The Complete Guide to Free Lossless Audio Codec

FLAC (Free Lossless Audio Codec) is the de facto standard for lossless audio compression. Developed by Josh Coalson and released by Xiph.Org Foundation in 2001, FLAC compresses audio without any information loss — a decoded FLAC file is bit-for-bit identical to its PCM source. With compression ratios typically between 40–60% of the original size, FLAC is the primary format for audiophile archival, CD ripping, and high-resolution audio distribution.

Why Lossless Matters

Lossy codecs (MP3, AAC, Vorbis) permanently discard audio information deemed inaudible by psychoacoustic models. Once encoded lossy, the discarded data is gone forever. FLAC stores every sample identically to the original:

Archives: Ripping a CD collection to FLAC creates a perfect digital archive. Re-encode to any lossy format later without generation loss.
Mastering and production: Audio engineers require lossless formats throughout the production chain.
High-resolution audio: 24-bit/96kHz, 24-bit/192kHz files from streaming services (Qobuz, Tidal HiFi) are delivered as FLAC.
Quality assurance: Blind A/B testing between FLAC and the original is statistically indistinguishable by definition.

FLAC File Structure

A FLAC file consists of:

1. fLaC marker: 4-byte magic number 0x66 0x4C 0x61 0x43 (fLaC)

2. METADATA_BLOCK sequence (variable count):

STREAMINFO (required, always first): 34 bytes — minimum/maximum block size, minimum/maximum frame size, sample rate (up to 655,350 Hz), number of channels (up to 8), bits per sample (4–32 bits), total samples, MD5 signature of the decoded audio stream
PADDING: Optional zero-filled block for future tag edits without file rewrite
APPLICATION: Embedded application-specific data with registered 4-byte ID
SEEKTABLE: Array of seek points (sample number → byte offset + samples in frame) enabling O(1) seek to any point
VORBIS_COMMENT: UTF-8 key=value metadata pairs (identical format to Ogg Vorbis comments). Common tags: TITLE, ARTIST, ALBUM, TRACKNUMBER, DATE, GENRE, REPLAYGAIN_TRACK_GAIN, REPLAYGAIN_TRACK_PEAK
CUESHEET: CD table of contents, track indices, ISRC codes
PICTURE: Embedded images (album art) with MIME type, width, height, color depth, color count, binary data

3. AUDIO FRAME sequence: Variable-size frames each encoding a block of samples

Audio Frame Internals

Each FLAC audio frame contains:

Frame header:

Sync code: 14 bits (0x3FFE) + reserved + blocking strategy
Block size: 4 bits (encoded or verbatim)
Sample rate: 4 bits
Channel assignment: 4 bits (independent, left/side, right/side, or mid/side stereo)
Sample size: 3 bits
UTF-8 encoded frame/sample number
CRC-8 of header bytes

Subframes (one per channel):

Subframe type determines the predictor:
- SUBFRAME_CONSTANT: All samples in block have same value (extreme silence)
- SUBFRAME_VERBATIM: Raw PCM, no prediction (useful for random noise)
- SUBFRAME_FIXED: Linear prediction with fixed coefficients (orders 0–4)
- SUBFRAME_LPC: Finite Impulse Response linear predictor with Burg-method or Levinson-Durbin computed coefficients (orders 1–32)

Frame footer: CRC-16 of entire frame

Linear Predictive Coding (LPC)

FLAC's compression core is LPC. Rather than storing raw samples, FLAC:

Predicts each sample from previous samples using a linear combination: predicted_n = sum(coefficients[i] * signal[n-i-1]) for i=0..order-1
Computes residuals: residual[n] = actual[n] - predicted[n]
Encodes residuals with Rice coding: residuals are typically small numbers (near zero) regardless of the original amplitude, making Rice codes efficient

The predictor order (1–32) trades accuracy vs. decoder complexity. Higher orders predict music better (more redundancy removed) at the cost of more coefficient storage.

Rice Coding (Golomb-Rice Entropy Coding)

FLAC uses Rice codes to compress the prediction residuals:

For each partition of residuals, a Rice parameter k is chosen that minimizes the encoded size
Residuals are mapped to non-negative integers (positive values → even, negative → odd)
Each value n is encoded as floor(n / 2^k) unary bits + n mod 2^k in binary
FLAC supports two Rice coding methods: Method 0 (parameters 0–14) and Method 1 (parameters 0–30)

The SEEKTABLE enables random access: without it, decoding would require reading from the beginning of the file (Rice-coded residuals are not independently decodable mid-stream without predictor state).

Compression Levels

FLAC offers 8 compression levels (0–8):

Level	Preset	Block Size	LPC Order	Compression	Speed
0	Fastest	1152	FIXED 0–4	~55%	Very fast
5	Default	4096	LPC 8	~50%	Moderate
8	Best	4096	LPC 12	~45%	Slow

The difference between level 0 and level 8 is typically only 10–15% additional compression at the cost of 10× slower encoding. Decoding speed is similar across all levels. Level 5 (default) is recommended for general use; level 8 for archival where encode time is acceptable.

Encoding and Decoding Commands

# Encode WAV to FLAC (default compression level 5)
flac input.wav -o output.flac

# Encode with maximum compression
flac -8 input.wav -o output.flac

# Decode FLAC back to WAV (verify losslessness)
flac -d input.flac -o output.wav

# Test integrity (decode in memory, verify MD5)
flac -t input.flac

# Encode with embedded album art
flac --picture="3||||cover.jpg" input.wav -o output.flac

# FFmpeg: convert to FLAC with 24-bit
ffmpeg -i input.wav -c:a flac -sample_fmt s32 output.flac

# FFmpeg: FLAC to MP3 V2 (transcoding from archive)
ffmpeg -i archive.flac -codec:a libmp3lame -q:a 2 output.mp3

# Batch rip CD tracks to FLAC with tags
for i in track*.wav; do
    flac -8 "$i" --tag=ARTIST="Artist Name" --tag=ALBUM="Album Title"
done

ReplayGain

FLAC natively supports ReplayGain metadata stored in VORBIS_COMMENT tags:

REPLAYGAIN_TRACK_GAIN: dB adjustment to normalize individual track loudness
REPLAYGAIN_TRACK_PEAK: Maximum sample value (0.0–1.0)
REPLAYGAIN_ALBUM_GAIN: dB adjustment for album-level normalization
REPLAYGAIN_ALBUM_PEAK: Maximum sample value across the album

ReplayGain targets -18 LUFS (older standard) or -23 LUFS (EBU R128 alignment). Applied by the player at playback time without modifying the audio data.

# Calculate and tag ReplayGain for an album
metaflac --add-replay-gain track01.flac track02.flac track03.flac

FLAC vs Other Lossless Formats

Format	Compression	Bit Depth	Channels	Container	Notes
FLAC	40–60%	4–32 bit	Up to 8	Native or Ogg	Open, royalty-free
ALAC	40–60%	16–32 bit	Up to 8	MP4/M4A	Apple format, iTunes/iOS native
WAV	0%	8–32 bit	Up to 18	RIFF	Universal, no metadata
AIFF	0%	8–32 bit	Up to 2	IFF	Apple legacy, uncompressed
WMA Lossless	50–60%	16–24 bit	Up to 8	ASF	Windows-only, DRM-capable
Wavpack	40–65%	8–32 bit	Up to 256	WV	Hybrid lossy+lossless mode
Monkey's Audio (APE)	50–70%	8–32 bit	Up to 2	APE	High compression, slow decode

FLAC's advantages over ALAC: Open specification, cross-platform (including Linux, Android), broader software support. ALAC's advantage: Native iOS/iPod/iTunes support without conversion.

High-Resolution Audio

FLAC supports high-resolution formats natively:

24-bit/96 kHz: Standard "Hi-Res Audio" from streaming services. File size ~5× larger than CD-quality FLAC (~60 MB for a 4-minute track).
24-bit/192 kHz: Studio master resolution. ~120 MB for a 4-minute track.
DSD-over-PCM (DoP): DSD64/DSD128 encoded into 24-bit PCM frames within FLAC for compatible DACs.

Sample rate support extends to 655,350 Hz, making FLAC suitable for ultrasonic research recordings.

Streaming and Seeking

The SEEKTABLE metadata block enables efficient random access — the player can jump to any time position by looking up the nearest seek point and decoding forward from that frame. Without a seek table, seeking requires linear scanning from the file start.

FLAC in Ogg container (OGG FLAC) uses Ogg page granule positions for seeking, compatible with Ogg streaming infrastructure. Native FLAC files are streamable but require the seek table for efficient seeking.

Related conversions

Common video conversions that pair well with this guide:

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, FLAC is mathematically lossless. The STREAMINFO metadata block stores an MD5 hash of the original decoded audio stream. Running `flac -t filename.flac` decodes the file in memory and verifies the decoded output matches the stored MD5 — if they match, the file is bit-perfect. You can also decode to WAV and compare with the original using a checksum tool.

Sim, FLAC is mathematically sem perdas. The STREAMINFO metadata block stores an MD5 hash of the original decoded audio stream. Running `flac -t filename.flac` decodes o arquivo in memory e verifies the decoded output matches the stored MD5 — if they match, o arquivo is bit-perfect. You can also decode to WAV e compare com the original usando a checksum tool.

Ja, FLAC is mathematically verlustfrei. The STREAMINFO metadata block stores an MD5 hash des original decoded audio stream. Running `flac -t filename.flac` decodes die Datei in memory und verifies the decoded output matches the stored MD5 — if they match, die Datei is bit-perfect. You can also decode to WAV und compare mit the original using a checksum tool.

Sí, FLAC is mathematically sin pérdidas. The STREAMINFO metadata block stores an MD5 hash del original decoded audio stream. Running `flac -t filename.flac` decodes el archivo in memory y verifies the decoded output matches the stored MD5 — if they match, el archivo is bit-perfect. You can also decode to WAV y compare con the original using a checksum tool.

AV1 is the most efficient (royalty-free, ~30% smaller than H.265) but encoding is slow. H.265 (HEVC) saves ~30–50% over H.264 and is supported by every modern phone and desktop. H.264 remains the safest baseline for legacy compatibility. Rule of thumb: archives → AV1, daily use → H.265, broadest reach → H.264.

The default level 5 is the best balance for most uses — good compression speed with solid file size reduction. Level 8 provides 10–15% smaller files but takes 10× longer to encode. Level 0 encodes very fast for batch operations where storage is plentiful. All levels produce identical decoded audio — only the file size and encode time differ. Decoding speed is nearly the same at all levels.

The default level 5 is the best balance para most uses — bom compressão speed com solid tamanho do arquivo reduction. Level 8 fornece 10–15% smaller arquivos mas takes 10× longer to encode. Level 0 encodes very fast para batch operations where storage is plentiful. All levels produce identical decoded audio — only the tamanho do arquivo e encode time differ. Decoding speed is nearly the same at all levels.

The default level 5 is the best balance für most uses — gut Komprimierung speed mit solid Dateigröße reduction. Level 8 bietet 10–15% smaller Dateien aber takes 10× longer to encode. Level 0 encodes very fast für batch operations where storage is plentiful. All levels produce identical decoded audio — only the Dateigröße und encode time differ. Decoding speed is nearly the same at all levels.

The default level 5 is the best balance para most uses — bueno compresión speed con solid tamaño de archivo reduction. Level 8 proporciona 10–15% smaller archivos pero takes 10× longer to encode. Level 0 encodes very fast para batch operations where storage is plentiful. All levels produce identical decoded audio — only the tamaño de archivo y encode time differ. Decoding speed is nearly the same at all levels.

CRF (Constant Rate Factor) is the best default for offline files: ffmpeg picks the bitrate frame-by-frame to maintain perceived quality. Two-pass is only better when you must hit an exact final size (DVD targets). Constant bitrate is for streaming with a fixed channel. For "smallest at quality X" always use CRF.

Both are lossless audio codecs with similar compression ratios (40–60%). FLAC is open-source and royalty-free with broad cross-platform support including Linux and Android. ALAC (Apple Lossless Audio Codec) is Apple's format, natively supported by iOS, iPod, and iTunes without conversion. Apple open-sourced ALAC in 2011. For Apple device users, ALAC is convenient; for everyone else, FLAC is the standard choice.

Both are sem perdas audio codecs com similar compressão ratios (40–60%). FLAC is open-source e royalty-free com broad cross-platform support including Linux e Android. ALAC (Apple sem perdas Audio codec) is Apple's formato, natively suportado por iOS, iPod, e iTunes sem conversion. Apple open-sourced ALAC in 2011. para Apple device users, ALAC is convenient; para everyone else, FLAC is the padrão choice.

Both are verlustfrei audio Codecs mit similar Komprimierung ratios (40–60%). FLAC is open-source und royalty-free mit broad cross-platform support including Linux und Android. ALAC (Apple verlustfrei Audio Codec) is Apple's Format, natively unterstützt by iOS, iPod, und iTunes ohne conversion. Apple open-sourced ALAC in 2011. für Apple device users, ALAC is convenient; für everyone else, FLAC is the Standard choice.

Both are sin pérdidas audio codecs con similar compresión ratios (40–60%). FLAC is open-source y royalty-free con broad cross-platform support including Linux y Android. ALAC (Apple sin pérdidas Audio codec) is Apple's formato, natively soportado by iOS, iPod, y iTunes sin conversion. Apple open-sourced ALAC in 2011. para Apple device users, ALAC is convenient; para everyone else, FLAC is the estándar choice.

Common causes: (1) variable framerate source rendered as constant (use <code>-vsync vfr</code> to preserve VFR); (2) different audio sample rates not resampled (add <code>-ar 48000</code>); (3) container limitations (MP4 with variable framerate misbehaves — prefer MKV during editing, encode to MP4 only at the end). Always run <code>ffprobe</code> on both source and output to compare timing.

A standard CD (74 minutes, 16-bit/44.1 kHz) compresses to approximately 200–350 MB in FLAC at level 5 (vs. ~650 MB uncompressed WAV). A 500-album collection would require roughly 100–175 GB. A 2 TB hard drive holds approximately 5,000–10,000 albums in FLAC, making large lossless archives practical on modern storage. For comparison, the same collection in MP3 V2 would be approximately 25–40 GB.

A Standard CD (74 minutes, 16-bit/44.1 kHz) compresses to approximately 200–350 MB in FLAC at level 5 (vs. ~650 MB uncompressed WAV). A 500-album collection would require roughly 100–175 GB. A 2 TB hard drive holds approximately 5,000–10,000 albums in FLAC, making large verlustfrei archives practical on modern storage. für comparison, the same collection in MP3 V2 would be approximately 25–40 GB.

A estándar CD (74 minutes, 16-bit/44.1 kHz) compresses to approximately 200–350 MB in FLAC at level 5 (vs. ~650 MB uncompressed WAV). A 500-album collection would require roughly 100–175 GB. A 2 TB hard drive holds approximately 5,000–10,000 albums in FLAC, making large sin pérdidas archives practical on moderno storage. para comparison, the same collection in MP3 V2 would be approximately 25–40 GB.

Yes if you only change the container: <code>ffmpeg -i in.mkv -c copy out.mp4</code>. This remuxes the stream without re-encoding, takes seconds even for hours of footage. Limitations: codec must be supported by the target container (e.g. you cannot put H.264 in WebM, only VP8/VP9/AV1). To shrink size you must re-encode.