quality as a function of time

I have been beavering away at my vinyl to digital project the past couple of weeks. There has been some noise about the quality of various digital audio formats. As much as we think CDs are wonderful, with the advent of the SuperAudio CD format, it begins to look more like an arms race. And that’s where it starts to take up a lot of time. (More here and here: note that in the Rosenberg piece, he calls CD audio — what some are using as the high-water mark — as a lossy format. At some point you have to accept that this starts to follow the model of Plato’s shadows on the cave wall: we never quite capture everything, despite all our technology and incompatible data formats. Sometimes it’s not the audio quality that defines the quality of the experience.)

I doubt I’ll be able to tell a lot of difference between a good recording in a less than perfect listening environment (how many of us listen in a tuned and acoustically adjusted music room?) and a not-quite-so-good one in a set of headphones (my Sony earbuds or my Sennheisers). And not all formats are available in all systems: I can use flac on my FreeBSD system but that won’t import into iTunes. It has it’s own “lossless” encoding algorithm but I can only use it on CDs. (Well, it occurs to me I could let it encode WAV files created from vinyl, but I’ve not tried that.)

One of the biggest hassles of all this is dealing with the files and their meaningless names when they’re encoded. As one learns in working through all this, the ID3 tags are where the playback system gets its information from: the human readable filename is for your convenience. So my workflow of digitizing in gramofile, splitting the album sides into tracks (going from 2 to as many as 14 files) and taking those anonymously named WAV files through an mp3 encoder and ending up with a bunch of slightly different anonymously named files gets to be a bit tedious and lacks any feeling of progress.

I had been using bladeenc with a bitrate of 192kbits: sounded OK. Then I looked at LAME and discovered that it can add ID3 tags info as it encodes, as well as make some decisions about bitrate to improve quality, on the fly. A little Google action and I can now save myself a bunch of time (and keystrokes) by tagging all but the track name at the same time as I encode the tracks. (If I thought this warranted it, I suppose some kind of cue-sheet/tracklist parser to handle track names could be devised, but I’m happy with the time savings this far.)

for i in processed*wav; do lame --preset extreme --add-id3v2 --pad-id3v2 --tl "Secondhand Daylight" --tg "Rock" --ta Magazine --ty "1979" $i ; done

Should be obvious, but I’m tagging all the soon-to-be-encoded mp3 files, encoded at rates up to 248k, with an artist, album title, genre, and release date.

The “–t*” tags are a Good Thing.

And since lame is available for OS X (through DarwinPorts), it should be possible to backfill tags on existing files without using iTunes’s GUI.