Initializing... drag & drop files here
Supports: MPEG2
The honest short answer first, because most pages about this conversion get it wrong: FLAC cannot restore anything your MPEG-2 file already threw away. FLAC is lossless with respect to what you hand it, not with respect to the original studio master. If the soundtrack inside your .mpeg2 is MP2 or AC-3 — which it almost always is, because that is what DVD-Video and DVB broadcast use — then converting to FLAC freezes that already-lossy audio perfectly and permanently. That is genuinely useful, and it is a different thing from "recovering" quality.
Where FLAC earns its place is as a stable, verifiable archive: every later copy, edit and re-encode starts from the same samples instead of stacking one lossy generation on top of another. And in the one case where an MPEG-2 file carries LPCM audio, the conversion really is a clean, complete transfer of the sample data.
| Property | Audio inside an MPEG-2 file | FLAC output |
|---|---|---|
| Compression | Lossy in almost every real case (MP2, AC-3, DTS); LPCM is the exception | Lossless — decoded samples are stored exactly |
| Specification | Defined by the DVD-Video and MPEG program-stream specs | RFC 9639, published December 2024 by the IETF |
| Typical bitrate | AC-3 up to 448 kbit/s; MP2 up to 912 kbit/s on DVD-Video | Content-dependent; commonly 400–1,000 kbit/s for stereo |
| Channels supported | Up to 5.1 on DVD-Video (7.1 for MP2 in spec) | 1 to 8 channels |
| Sample rates | 48 kHz on DVD-Video (96 kHz for LPCM/DTS) | 1 Hz to 1,048,575 Hz |
| Bit depth | 16 or 24 bit for LPCM; lossy codecs do not store one | 4 to 32 bits per sample |
| Editable without loss | No — every save re-compresses | Yes — re-saving does not degrade |
| Plays on phones by default | Yes, as part of the video | Android yes; iOS via Files/VLC; every desktop player |
| Good for | Playback of the original disc or broadcast | Archiving, editing, mastering, re-encoding later |
.mpg, .mpeg or .mpeg2 file onto the page or click "Add Files." Several files can be queued and processed with the same settings.An MPEG program stream is a wrapper; the audio inside it can be any of four things, and which one you have decides what FLAC is worth to you.
| Source track | Where it comes from | What FLAC gives you |
|---|---|---|
| MP2 (MPEG-1 Audio Layer II) | The default DVD-Video track in PAL regions, and most DVB broadcast | A permanent, exact copy of the lossy decode — no further loss from here on |
| AC-3 (Dolby Digital) | The default DVD-Video track in NTSC regions, up to 448 kbit/s | Same: the decoded result is frozen losslessly, but the Dolby metadata and channel layout are not a FLAC concept |
| DTS | Optional DVD-Video track at 754.5 or 1509.75 kbit/s | Same as AC-3 — a faithful copy of the decode |
| LPCM | Uncompressed DVD-Video audio, 48 or 96 kHz at 16 or 24 bit | A genuinely complete transfer: FLAC stores those samples exactly, typically at about half the size |
Note that only one audio track survives the conversion, and there is no track picker on this page. Selection is automatic and follows ffmpeg's documented rule: the audio stream with the most channels wins, and ties are broken by the lowest stream index. On a disc that carries a 5.1 AC-3 track alongside a stereo commentary, the 5.1 track is the one you will get.
No. Nothing can rebuild detail that a lossy encoder already discarded, and the MP2 or AC-3 track inside a DVD-era MPEG-2 file is lossy. What FLAC does is stop the erosion: from the moment you have the FLAC, no further generation loss accumulates no matter how many times you edit, copy or re-encode it. Treat the conversion as preservation, not restoration — any tool that promises "pure, unaltered sound" from a lossy source is overselling it.
File size and encoding time, and nothing else. FLAC is lossless at every setting, so level 1 and level 12 decode to identical samples — level 12 simply spends more CPU searching for a tighter prediction model. We ship the slider at 12, the top of its range, where ffmpeg's own FLAC encoder defaults to 5; drop it if you are converting a long recording and would rather have the file sooner than have it a few percent smaller.
Usually several times the size of the lossy track it came from, because you are storing a lossless representation of a lossy decode. A 192 kbit/s MP2 track expands to whatever FLAC needs for that waveform, commonly in the 400–1,000 kbit/s range for stereo music and lower for speech. Against uncompressed audio the picture flips: FLAC typically lands near half the size of the equivalent WAV.
FLAC itself supports 1 to 8 channels, so a multichannel result is possible in principle, and automatic stream selection will pick your 5.1 track over a stereo one. What you should not expect is Dolby Digital's metadata — dialogue normalisation, downmix coefficients, dynamic-range control — to survive, because those are AC-3 features with no FLAC equivalent. If you want a guaranteed two-channel result, set Audio Channel to STEREO before converting.
Leave Audio Sample Rate on ORIGINAL. DVD-Video audio is 48 kHz (96 kHz for some LPCM and DTS tracks), and resampling it to 44100 Hz to look more like a CD is a lossy arithmetic step that gains you nothing — FLAC is perfectly happy storing 48 kHz. The lower rates in the list (8000 to 24000 Hz) exist for speech-sized files and will audibly dull music.
Because there is no track selector here and the automatic rule is channel count, not language. A disc with an English 5.1 track and a French stereo track will give you the English one; a disc where the commentary is the 5.1 track will give you the commentary. If the track you want is not the one with the most channels, demux it in a desktop tool before uploading.
Yes — use Trim on this page, which takes a Start time and a Duration and applies before encoding, so you are not waiting on a two-hour file to convert. If you would rather cut the video first and convert afterwards, the video cutter accepts MPEG-2 directly. The audio cutter does not take video files, so trimming has to happen either here or on the video side.
Both. FLAC came out of the Xiph.Org Foundation as free software, and in December 2024 the IETF published it as RFC 9639, which specifies the bitstream formally: 1 to 8 channels, sample rates from 1 Hz to 1,048,575 Hz, and 4 to 32 bits per sample. That matters for archiving, because a documented, royalty-free format with multiple independent decoders is exactly what you want a long-term master to be stored in.
It is uploaded over an encrypted connection, processed on our servers, and deleted automatically after a few hours, along with the FLAC we generated. There is no sign-up and no watermark. Since a two-hour MPEG-2 rip can be several gigabytes, the slow part of this job is usually the upload rather than the encode.