By Raj Jayavant · Updated
A lead sheet for yourself and a note-for-note transcription for publication are different jobs. Deciding which you are doing prevents a great deal of wasted effort.
For most practical purposes, chords and melody with approximate rhythm is enough, and takes a fraction of the time. Reserve full detail for material where the articulation is the point, like a solo you are studying.
Listen through and write down the sections and their lengths: intro eight bars, verse sixteen, chorus eight. Do this before transcribing anything.
You will usually find large parts of the piece repeat exactly, which means the actual transcription work is much smaller than the duration of the track suggests. It also stops you from discovering in hour three that the second verse was identical to the first.
Work out the bass note of each bar, then whether the chord above it is major, minor, or something else. This gives you a frame that constrains every melodic decision afterwards.
The approach is the same one described in how to learn a song by ear, which covers finding the key and working in short sections.
First pass, get the pitches. Second pass, fix the rhythm. Trying to capture both at once is where transcription usually stalls, because each one requires a different kind of attention.
For rhythm, count out loud along with the recording while it plays slowly. Rhythms that are ambiguous at tempo are usually obvious at half speed, particularly the difference between a triplet and a dotted figure.
This is the whole technical difficulty of transcribing, and it is solved. A fast run contains notes that are perfectly identifiable at half or quarter speed, and modern playback preserves the pitch when you slow it, so what you hear stays in the original key.
The two things that make this workable in practice are looping a short section so it repeats while you work, and being able to change speed without losing your place. PhraseLoop does both, and works with a YouTube link or an audio file from your own machine, which matters when the recording you are transcribing is not online.
Play your transcription against the recording at reduced speed. Errors that are invisible on paper are obvious when your version and the original sound simultaneously.
Do this section by section as you go rather than at the end. A mistake caught immediately costs a minute; the same mistake found after you have written the next thirty bars around it costs considerably more.
Real performances are full of small deviations: notes slightly ahead of the beat, a bend that does not quite reach, an implied rather than stated rhythm. Transcribing every one produces something unreadable.
Write what you hear first, then simplify deliberately for whoever will read it. The simplification is an editorial decision, and it is better made knowingly than by failing to hear the detail.