Uploading a file

Not every record starts at a turntable. A purchased download, or audio captured on another machine, can be brought in from the browser and cut exactly like a recorded side.

Doing it

1. Open the Upload tab.
2. Leave the record selector on — new album — and type the Artist and Album, or pick an existing record. The names work the same as they do in the Rip panel, because they are the same fields.
3. Choose a file… The right-hand column names the file, its size, and what it will become.
4. Upload. The bar fills and the line under it counts up.
5. When it finishes, the panel says wrote side-a.flac — 42:18, 287 MB and the record opens with its waveform built.
6. Go to Cut and carry on as normal — first pass, check the boundaries, cut, import, archive.

The mode does not change by itself. A finished upload leaves you where a finished rip leaves you: the record is ready and you choose when to move on.

What it accepts

Anything ffmpeg can read — FLAC, WAV, MP3, AAC, ALAC, Opus, and the audio track of a video file.

Stereo or mono. A file with more than two channels is refused, because a six-channel track in a music library is a problem discovered late.

Up to 2 GB.

What happens to the file

It is always re-encoded to FLAC, even if it already is one. A file named .flac may be FLAC in a different container, or at a rate and depth of its own, or carrying an embedded cover. One pass makes it the same shape as a recorded side.

Two things are deliberately dropped on the way in:

• The cover image, if there is one. It would otherwise become a second stream and make the side's duration read as zero.
• All tags. A download arrives tagged, and those tags are exactly what the first pass and beets are about to decide. A recorded side carries none, so dropping them makes the two identical.

Nothing is resampled and nothing is remixed. The rate and depth that arrived are the rate and depth that land.

A lossy source stays lossy. Encoding a decoded MP3 to FLAC is lossless from that point onwards, which is the best available — it does not recover anything.

The file is checked before and after: it is refused if it has no audio in it, and the encoded side must decode end to end before it is placed. Nothing is ever written under a side name until it is whole, so a half-arrived upload can never be mistaken for a record.

If it is interrupted

The part that arrived is kept. Come back to the Upload tab and it says how much is there:

album.flac — 142 MB of 613 MB already here. Pick the same file again to carry on.

Choose the same file and press Upload; it continues from where it stopped rather than starting again.

You do have to pick the file again. A browser cannot re-read a file you chose before the page reloaded — that is a rule of the browser, not a limitation here. The name and size are checked against what is already there.

Cancel discards what arrived. If the record has nothing else to it, the name goes with it.

Uploads nobody comes back to are cleared out after a day.

Where it lands

The file becomes side-a.flac of the record, in raw/<slug>/ — the same place a recorded side goes. See Files on disk.

One file per record, holding the whole album. If a record already has a side a, the upload is refused unless you mean to replace it.

A thing worth knowing before you cut

A gapless album will not fit well. The boundary detection looks for quiet stretches near where the catalogue says a track ends. A CD-sourced download often has none — the tracks run into each other — so the first pass falls back to the catalogue's durations with nothing to snap to.

The delta column will be honest about it, and every boundary is still correctable by ear. But it is more hand work than a record with real gaps between tracks.

Also worth asking first: a purchased download is usually already one file per track. If that is what you have, you do not need RipDoctor at all — hand those files straight to beets. This path is for one long file that needs splitting.

From the command line

There is no upload command. The equivalent is to put the file in place yourself and name it, which is described in Ripping a side — though doing it that way skips the normalising and the checks above.