This is how I have learned on my own how to structure my transcription workflow.
- Find the track on YouTube and work with this as your source. I haven’t yet found anything that I wanted to transcribe that isn’t there.
- If you need stem splitting, then search “youtube to mp3” on the internet, and convert the YouTube URL to an mp3 file on your computer. Then you can drop the mp3 into an AI stem splitter. I have found that stem splitting (a) doesn’t separate horns from each other, but separating them from other instruments can help with hearing the harmonies, (b) isn’t consistent with keeping all the horns in the same stem – some parts or sections might end up in “other”, and (c) is especially useful for transcribing bass, which is of course helpful for establishing a backbone for harmony. If I am transcribing a single soloist over a rhythm section, I have not found stem splitting to be useful. Moises is the stem splitter I have started using, and it does have variable tempo in playback, but when I slowed a track down below half speed, let’s say, the pitch did drop.
- YouTube also has variable playback speed and does preserve pitch. 0.25 is usually too slow for discerning rhythm, so I mostly work with 0.5; I’ll go down to 0.25 if the rhythm is already transcribed and I just need to have a closer listen to a few pitches.
- When working with the notation file itself (I use flat.io), I often remind myself to create the blank choruses first before beginning to transcribe. So, listen to the whole solo to figure out how many choruses it is. Label the beginning of each chorus in the transcription with the chorus number and the time stamp. Include the chords in the transcription of a single blank chorus so that when that chorus is copied and pasted, it brings the chords with it. Put a rehearsal marker (e.g., “A”) at the beginning of each chorus. Put a double bar at the end of each chorus.
- Learn the shortcut keys to reduce clicking, which can be annoying. Do whatever you can to make the task enjoyable and not a chore. Maybe transcribe only one chorus per day.
- I’m a little conflicted about whether to include any small errors on the part of the soloist. Sometimes you can tell that the soloist was executing a repeating, transposed pattern and one of the notes doesn’t fit the pattern. Or maybe the pattern is syncopated and the soloist doesn’t quite nail the time. If I feel confident that I know what the soloist intended, and the figure goes by relatively quickly, then I will transcribe what I think the intended figure is, in case I can possibly get it as I play it. Odds are when I try to play it, it will come out imperfectly as well, because there is some systematic, instrument-based reason that the figure is difficult. So this may be the best way to make it sound like the original anyway.
- Transcribe rhythm first if you can’t do rhythm and pitch concurrently. Mark all articulations and ornamentations after all the rhythms and pitches are in place.
- Include in the transcription information (a) your name! I have found transcriptions that don’t say who did it. I even found one that said it was transcribed by “me”. I usually put this in the “Arranger” field, like “tr. [transcriber’s name]”; (b) the original composers of the tune; (c) identifying information for the performance in the subtitle: soloist’s name, name of album, original year released. Go to discogs.com to find this information if you don’t know it. Don’t honor release dates from compilations of earlier works.
- I began my journey with transcription by finding YouTube videos that already included transcriptions shown one system at a time while the track played. I would copy what was there into my own notation file. I don’t do this anymore. At first, I thought I should hide that I was “cheating” by looking at others’ transcriptions, but now, I feel fine about that as a developmental step in the process of learning transcription. If that’s where you are, I think it’s fine, as long as you see it as a stage to challenge yourself to grow out of after a few rounds.