Three Weeks Out From a Support Slot, What Actually Helps

The honest answer to “do these music tools do anything for a working artist” is mostly no, and I say that as someone who has paid for a depressing number of them.

Most arrive promising to write your song. Nobody I know wants that. What people want, three weeks before a support slot with four covers to learn and no budget, is duller and much more specific: hear a part properly, get it into their hands, stop losing the idea they had on the bus.

So here is one preparation, start to finish, with the parts that helped and the parts that did not.

Three weeks out: the part you cannot hear properly

The set has a soul number in it. There is a bass figure under the second verse that everybody in the room hears differently, which is a polite way of saying nobody has worked it out.

Slowing the record down does not fix this. The bass is buried under a vocal that sits in roughly the same frequency neighbourhood, and slowing it down buries it more artistically.

Taking the voice off the record first is the move. A vocal remover online will hand back an instrumental in a couple of minutes, in a browser, with no account until you want to download. Suddenly the figure is sitting there in the open, and the argument ends.

“It stopped being three people guessing,” a bassist I have worked with put it. “It became one person transcribing, which is a job that finishes.”

Worth being clear about scope, because a vocal remover is sold as more magic than it is. It takes a finished stereo mix and gets the lead voice out of the way. It is not going to rebalance the horns, and on a dense 1970s master you will hear a bit of ghosting where the voice used to be. Fine for learning a part. Not a released stem.

Two weeks out: turning a part into something you can read

Here is the bit I did not expect to use twice a month.

That instrumental you just made does not have to be re-uploaded anywhere. It sits in your history, and mp3 to midi will pull it straight through from there, along with anything else you have generated on the site. Note detection runs on Basic Pitch, and what lands at the end is an ordinary MIDI file that any DAW will open.

It takes the awkward formats too, incidentally, which the vocal remover does not. FLAC and AAC both go in, so the lossless copy of the record you keep for exactly this sort of thing does not need converting first.

You get the detected notes on screen before you commit to anything, previewed on a grand piano, which is unglamorous and exactly right. A bass figure previewed on a piano is legible in a way it is not as a waveform. You can see the passing note you have been playing wrong for a fortnight.

Two habits, learned the tedious way. Check the octave, because machine transcription is confident and frequently an octave out. And keep expressive pitch movement in mind: slides and bends survive the conversion rather than getting tidied away, which is honest to the record and slightly startling on a piano roll.

Ten days out: the thing on your phone at 1am

Everyone has the voice memo. Half a chorus, hummed badly, over a bus engine.

The microphone route takes it directly, five minutes a take, and puts the shape of the thing onto a grid where it stops being weather and starts being notes. Is that worth doing? Depends how you write. Songwriters who work at an instrument mostly do not need it. Songwriters who work by singing, and who have lost a good one before, know exactly why it matters.

“I have got about forty memos I will never open again,” a songwriter told me, cheerfully. “The four I bothered to convert are the four that became anything.”

The night before: what still goes wrong

Your instrumental is on your phone and it plays through headphones beautifully. It will not sound like that through a small venue’s monitor wedge.

Play it out loud, in a room, at something like performance volume, at least once before you rely on it. What a vocal remover leaves behind is the mix minus a voice, and any artefacts hiding under the vocal in the original will be sitting exposed in the space where the vocal was. Usually inaudible. Occasionally a swirl you cannot unhear once the PA is loud enough.

Have the original with you regardless. The number of times a stage manager has asked for the actual track, and the answer was “I have got a version with the singing taken out”, is a number I would prefer not to give.

When to shut the laptop

Two situations, both of which I have got wrong.

The first is anything an audience will hear as a finished piece of work. A vocal remover is a rehearsal tool and a transcription aid. If the backing you use on stage is a stripped commercial record, that is a licensing conversation with the promoter, and it is a conversation to have well before the day.

Second failure: converting instead of playing. There is a version of this week where you spend eleven hours making beautiful MIDI files of parts you could have learned by ear in ninety minutes. The tools are quick enough now that this failure mode is genuinely easy to walk into.

My rule for this is unromantic. Use a vocal remover when a part is buried and the room disagrees about it. Use the converter when something needs writing down or would otherwise be forgotten. Anything else is procrastination with a good user interface.

Learn the songs.

Related Posts