Showing posts with label Pinned. Show all posts
Showing posts with label Pinned. Show all posts

Monday, 29 September 2025

Steve Slate VSX Listening Routine

What to Listen For, What to Check, How Do You Use These Things Anyway?

This is a reminder to myself about how to do a mix translation test using VSX headphones through the VST plugin, after doing the bulk of mixing in Linear Mode.

* * *

1. Linear (Flat / Microscope View)

Play 20–30 sec mid-strength section

  • Check local loudness relative to snare/kick.
  • EQ balance has no harsh spikes, not dull anywhere.
  • Bass and kick drum has nice separation.
  • Listen for sibilance, cymbal/hi-hat/high end harshness
  • Mono check (quick flip: does vocal stay centred? Does low end collapse?).

2. Mix Room – Nearfield (Producer’s Laptop Monitors)

Play the same section.

  • Vocal placement sits in the mix, not on top or buried.
  • Snare/guitar/keys clarity in the mids.
  • Kick vs bass relationship in low end okay?
  • Does reverb delay stick out?
  • Small level tweaks (1-2 dB) often shine here.

3. Mix Room – Midfield (Full Mix Perspective)

Play peak song section.

  • In low end does kick + bass still feel solid.
  • Panning/reverb/stereo balance feels okay, no phaseyness.
  • Automations and arrangement flow - do transitions lift the energy?
  • Overall size - not too thin, not boxy.

4. SUV (Car Test)

Play full chorus/bridge at a higher volume.

  • Vocals still audible? (cars bury them if too low).
  • Kick + bass groove not too boomy?
  • Sub-bass control - does anything rattle or overwhelm?
  • Harshness at loud playback - does the mix stay smooth?
  • Does the song feel exciting/hyped?

5. SA-Pods (Earbud Reality Check)

Play chorus at low–moderate volume.

  • Vocals cut clearly (lyrics understandable).
  • Snare/guitar mids provide drive even if bass vanishes.
  • Quiet-volume test - does the mix still feel compelling?
  • If it works here, it works anywhere.

Sunday, 25 August 2024

My Reaper Mixing Notes

This article is a work in progress and will be constantly updated, for me, as I learn new skills!

I use the Digital Audio Workstation (DAW) called "Reaper" for mixing and I quite often get a multi-track recording from a Behringer X32/Midas M32 deck (they are virtually the same thing). This post is a dumping ground for all the things I have learnt from the process so that I have a coherent centralised place to keep them for me - and maybe they will also be useful to you.

Remember some golden rules: use your ears, not your eyes; keep it simple; mix in context not in solo.

Important things to do to Reaper before you start:

  • Download and install SWS Extensions, it's, umm, to enable you to, ahh... just do it. sws-extension.org
  • Download and install ReaPack, for managing scripts. Reapack.com
  • Options->Preferences->Project Backups and tick the top level Auto-save. To prevent heartache later on when you forgot to press save when you should have, or presses save when you shouldn't have.
  • File->Project Settings:
    • Set you desired sample rate - 48kHz for an X32/M32 deck
    • If you are working to a click track, set the BPM here.
    • If you are not working to a click track, set "Timebase for items/envelopes/markers" to "Time" (won't automatically stretch items pasted from a different timebase)
  • In Reaper use the WASAPI driver instead of Windows direct sound, it is better. Options->Preferences->Device and select WASAPI. Go for Shared Mode so that other apps can use sound. Set the sample and bit rate to say 48kHz 24 bit. In Windows sound settings make sure you input and output devices the same setting. If you get the error "WASAPI shared mode device error: input device sample rate does not match output device samplerate!" then hard select the Input device and Output device to match your windows devices, you shouldn't need to fall back to exclusive mode.
  • You can save fx chains and reuse them, very handy.  They MUST be in the Reaper FXChains directory so be seen in the fx dialog, so any that you have (or save) will be (Windows) in C:\Users\<you>\AppData\Roaming\REAPER\FXChains
  • Windows users: In System->Sound your device->Properties MAKE SURE AUDIO ENHANCEMENTS ARE OFF! And Spatial sound. And anything that Windows uses to mess with the signal.

Raw files

Acquiring from X32/M32

While the X32/M32 has a USB port on the front it will only record a deck-mixed stereo signal. However if you have the SD card module added, you can record all 32 channels pre-mixer. Format any SD card to FAT32 and pop it in. While the manual talks about 32GB as the limit, I have been using 128GB cards no problem which can record a 5 hour session. Class 4 SD cards are fine for bandwidth.

It records everything into a WAV file, split into 4GB chunks, not useful yet. Here is my process:

  1. They are hexadecimally named which doesn't sort well, so I made this little script to rename them:Rename x32 hex to dec.cmd.
  2. Drag all these files directly into a Reaper track, "sequential time positions on a single track". You will now have a single track containing multiple items (the wav chunks), each containing the 32 channels.
  3. In Reaper, Extensions (previously installed from ReaPack) browse and install the package "Explode multichannel items to mono items in new tracks (without render)" by Rodilab. Run the script as an action. It will explode the WAV into separate tracks.
  4. In Reaper run my "prep" script which will get rid of the unused tracks, add colour, group, pan, based on rules. Download this script and edit it according to the rules you need. This will require some experimenting on your behalf! JAW X32 Track Rules Parser.lua

Consolidating

Now that you have multiple mono tracks with multiple items, all referencing rather large WAV files, it's a good time to consolodate into smaller MP3 files and throw away the unused tracks and quiet sections.

  1. Delete any tracks that weren't active, eg, there was no electric guitar playing that day, only 3 vocalists, etc)
  2. Set Reaper into "Ripple Editing All Tracks" - there is a button on the toolbar. Click to show a cursor, ctrl-A to select all tracks, S to split. Then use right mouse to drag click select, press delete and sections of the tracks will be removed. Turn off Ripple Editing
  3. File->Consolidate/Export Tracks, create a new directory, give it a name, set the resample mode to the highest quality MP3 or FLAC if disk space is not an issue, entire project all tracks, and tick all the boxes to update project with consolidated files and saving as a new project. This will create a much smaller set of files but still high quality files, with only one item per track. Much easier to deal with. Will take quite a while, all tracks will be rendered.
  4. Ctrl-A, Right click on any item, Item Processing, Normalise items. Normalise to Peak, "Normalise each item Separately". Depending on what the gains were like, this will get all tracks to the highest amplitude without any distortion. Again, to make it easier to deal with.
    NOTE: for dual mono channels that represent a stereo signal, do these "Normalise to loudest item" to ensure that they remain "paired" as part of the normalisation process. Otherwise you risk making one channel louder than the other.

Rough Mix

Faders(volume)

Set all faders to an appropriate level before applying any effects to get a rough mix. GIVE YOURSELF SOME HEADROOM! Don't try to max out the gain at this stage, leave yourself some room in the master track, don't worry, you can make it loud later.

You should only add effects to fix problems. With that said, if you already know a heap of problems that happen every time, and you have already solved them before, then add those effects in.

Pan

Not really an effect as such but definitely think about where you want each track to be situated from left to right. It's a good idea to get tracks off centre, even just by 10%, to create 3D space, unless you specifically want your track to be straight down the middle. With that said, to really create width pan tracks 100% left or right. Particularly good for multiple vocalists or a solo instrument that you can to hear "over there". Don't be frightened about going to extremes.

Phase

Maybe some of your mics have a different latency, and maybe the latency will cause phase cancellation or distortion. but if you want the tightest of the tight, then zoom right in on your master track and see how all the tracks are overlapping, specifically on punchy transients. Particularly useful on the drums, if you notice some tracks are offset to each other, you can move tracks in time slightly to line up the signals.

In Reaper 7.40 there is now an action to auto-align phase. Run the action "Item edit: Phase alignment" and let it do the work for you! Tip: For really long items, select a small section, tick "Use time selection" and the calculations will be much quicker. Always let it invert phase - it knows best. Cut a bit off the front of the tracks to allow the system to move tracks left, and hit apply.

Double Tracking

If you have recorded more than one track of the same song, to a click track, then add in a new track, synchronise them, hard pan one track left and one track right. This is fantastic for acoustic guitars. You might need to go through and re-align any errors that stand out - but use your ears, not your eyes. Perfectionism is a killer.

Routing Tips

When you create folders, containing children tracks, THE CHILDREN ALWAYS ROUTE TO THE PARENT. So if you wanted to route from one track to another, they can't be related. So for example say you had a complicated chopped up vocal track and you wanted a clean version with several vocal effects, but then another grungey version with abunch of unrelated effects, you would have three tracks all at the same level, you would then drag the routing icon from the first chopped up track to the clean track, then again to the grunge track. Then you can add you effects to those other two tracks.

Make sure you click on the routing for the sending track and untick Parent Send channels so that track isn't going to the master. Note that that track will act like a parent folder - if you solo it you will here all the tracks it is sending to coming back through it.

Vocal Isolation

Meta (Facebook) have been working on AI in the music scene for a while now, and to show how good they are they release their models for free! If you have a crowd mic and you want to incorporate it (it really adds ambience to a mix!) but it is full of reflections of the PA you can get quite an impressive vocal isolation, albeit not perfect. It will have artifacts which if glued nicely at a reasonable level in the mix won't be noticed. You can do this locally on your own PC, don't waste time with online isolators:

Initial Setup:

  • install python, make sure it is in your environment path. If you don't know how to do this, chat to the LLMs.
  • download the program "ffmpeg", unzip into a directory, add ffmeg/bin to environment path. If this doesn't make sense, chat to the LLMs.
  • in a command box execute:
    • pip install demucs
    • pip install soundfile
  • Do some research with your favourite Chatbot about GPUs.  Install the latest driver, look up how to use Torch with your GPU to really speed the process up.  Something like:
    • pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu124

Ongoing Use:

  • turn your crowd mic into an mp3 - if you have a stage left and stage right then hard pan them and save as a stereo mp3.
  • in a command box run: demucs -n htdemucs_ft "your audio file.mp3" --two-stems=vocals --device=cuda
    Note: htdemucs_ft is slower but better.  For faster response just use hrdemucs
    Note: drop the --device=cuda if you don't have a GPU set up
    Note: you might need to set the GPU if you have more than one:
       python: device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu")
       bash: set CUDA_VISIBLE_DEVICES=0

Effects

Delay

My favourite use of delay is a stereo delay - having different times in the left/right channels. This can make a mono signal sound like a wide stereo signal and is great to re-situate a track into a different 3D space to the rest of the tracks. Use "ReaDelay", set length to say time=80ms (musical=0), pan full left, then add a new tap, set to time=120ms pan full right. Adjust the wet for a good feel. Adjust the times to get the spacing you like the sound of.

Compression

Tame a signal where the dynamics are bigger than you want them to be, squeezes the levels together. Use on almost anything. Set the ratio higher to have less effect, but 4:1 is a good starting point. Move the threshold up - if you aren't exceeding the threshold anywhere, you aren't going into compression, there is no point having it. Set the attack fast if you need to the compressor to respond quickly, set the release slow if you want the compressor to not come off so quickly, sounds more natural. ReaComp is a good go to, don't forget to tick "Auto make-up" to keep the volume consistent. Choosing one of the stock presets for your application is a great starting point.

The JS 1175 is an emulated Urei 1176 fast compressor worth looking at, ReaXComp get even more fancy by having multiple frequency bands of compression.

Limiter

When compression isn't enough...use a limiter to prevent clipping on aggressive transients while bringing up all the subtleties underneath. Great on drums and in mastering to Make It Loud.

Set attack and release times so it doesn't sound unnatural.

For a The Ultimate Limiter use "Event Horizon Limiter/Clipper" which has zero delay response, this will clean up anything that escapes out of the normal limiter.

EQ

Boost and reduce frequency ranges, surely everyone knows what EQ does. A good way to work out what frequencies you want more or less of is to make a fairly narrow band boost, drag it extremely high, then sweep across the frequency ranges listening for what you like and don't like, and set it accordingly. Some good rules of thumb:

  • 20-70 : Rumble/Subbass : for dums and bass need this.
  • 70-300 : Boomy/Warmth/Muddiness/Fullness : Sometimes there are resonances here, notch them out.
  • 300-1.2k : Boxy/Nasal : often worth reducing on some tracks to create space for other tracks.
  • 1.2k-5k : Honky/Harsh : can add clarity. Look to remove vocal esses (sibilance) in here.
  • 5k-11k : Sparkle/Presence
  • 11k+ : Air

Use ReaEQ for creating normal boost/cut EQ. Use ReaFIR set to Dynamic EQ mode which can be great to tame harshness in vocals or solo instruments: Set mode to Compressor, set the Compression Ratio fairly low, say 1.5:1 and then set up a band profile where the compressor is starting to do stuff on the peaks across the frequency spectrum. This will smooth out peaks in frequencies depending on how hard you set the compression ratio.

Reverberation

Ubiquitous and overused but quite essential. Because we aren't always recording in a warm room or a cathedral, but those sounds are nice, so apply artificial versions of them. Don't overdo it! Argh! Reverb does not make up for a bad recording! A little bit of reverb underneath a vocal or a solo instrument can be quite nice, a good one as recommended by Kenny Gioia is ReaVerb: set to file, use the "Fat Plate" impulse file from the Lexicon 480L last seen at http://www.housecallfm.com/download.php?f=BGLex480.zip somewhere around -30dB.

Another great free reverb is Magic7.

A simpler Reaper built in reverb that can be quite effective is ReaVerbate: set the room size to massive - even say 95, and use very sparingly to make a vocal ring. Use side chained or automate the wet mix or manually add it in a separate track.

Gate

If something sits on background noise most of the time and then bursts into life, then gate it - ie, it will be silent unless the gate is exceeded. It will sound unnatural solo'd, but in a mix it will be fine. Drums are good to gate, sometimes even the vocals. Use ReaGate and set the attack/hold/release to something that suites your track, and most importantly set the threshold.

Tuning

It's okay to tune up a few vocals here or there, especially when there are multiple singers doing harmonies, but you can't fix a bad recording. Add in ReaTune in Auto mode, if you choose the key for the song it will make sure no notes leave the key - or just set it for chromatic so any normal note is allowed - so any off notes will snap to the closest note. 

If you need to fix just a few bad vocal notes is to do it manually: on the Correction tab change the attack time to around 100ms, then on the Manual Correction tab tick "Manual Correction" and then draw blue lines to fix missed pitches. Put your blue line slightly after the point at which the pitch is hit, to make it sound smoother, and finish the line a bit early. Put ReaTune first in your effects chains. Note - if you have to manually tune notes, it's probably already a lost cause...

Sometimes when there is sibilance ('S'es/etc) the tuner will go warbly and yuck, you will notice it. A straightforward way to deal with it is to simply turn off the tuner during those sections. Go to ReaTune, click Param and select "Show Track Envelope" (use "Wet").  Now you have an envelope under the track, simply block out the sections where you are hearing tuning nonsense. Using "Wet" to turn it off is a better idea than "Bypass" as you can often get pops and clicks and the VST energises - when you are using wet it continues to run in the background, you just don't use it for a bit.

Saturation

This is a quirky little effect, basically distorts the signal based on the way old analogue equipment like tape machines or tube amplifiers would distort when driven hard. Adds harmonics which are "richer, warmer" ie pleasing. I found the free plugin "GSat+" does a good job, load it up and at the top menu press the right arrow to say "Medium" or "Warm", wind the input up until it is going onto saturation for at least part of the range.

Another good saturator is Korneff "Chocolate Milk" which is a tape based saturator rather than the GSatplus tube based saturator. Because saturators are simulations of the real word effect of overdriving analogue signals through something like a vacuum tube or magnetic tape.

Quantisation

With your drum MIDI that you have captured and had played to a click track, and you want to fix up mis-timed hits, first start with quantisation:

  • Double click your MIDI to get the MIDI editor window
  • In the bottom left, first choose your grid - if the fastest notes you played for example were 1/16 then set it to that. Change to straight/swing, or triplets if you were playing them.
  • Press Q to bring up the quantise window.
  • A simple quantise would be: Use Grid, Uncheck Bypass, All notes, Position, Strength around 80% so it still sounds like a human played it and Fix overlaps, Hit commit and OK.
  • You might like to get your snare snappy and accurate - select all the snare notes and hit quantize again, set them to 100%
  • Now listen, and anything that was too out of time will be pretty obvious, click and drag into the right position.

Tip: when playing with midi, make a text file where each line is <midi note> <description> and import in with File - Note/CC names - Load note from file. Set the view to be note names rather than piano.

Dealing with Takes

If you are layering/overdubbing, Reaper will automatically make takes into a second lane in that track. You can be a total pro about it and comp it and finalise a take, or you can explode it into individual tracks and do it olde school. This is probably essential if you "accidentally" recorded more than just one thing into a single track. Right mouse drag to select all items in track, right click on one of them, Take - Explode all takes into tracks. For bonus points you can re-select all split items and right click - Heal items in split, as doing takes will naturally split at the point your take started and ended.

Instruments

Drums

The setup I spend a lot of time with has 7 mics. I will always make a track that is a folder (bus) for all the drum tracks and I add a limiter to that to "clean up" those times when the drummer has hit 3 drums full tilt at the same time.

  • Gate all drums except the overheads. Set the gate so that only the "fundamental" drum sound comes through, so usually the fastest attack possible and only a hold/release for as long as required, hopefully not before the next hit.
  • Put on a limiter rather than a compressor. Drums have a massive impulse on the initial hit which doesn't suit any compressor I've played with. Whereas the limiter, with a brickwall ceiling, can bring up the tail without losing too much initial impulse. Push the threshold until the limiter is only just coming in.
  • Add some saturation where you need a drum to really punch
  • EQ to the drum in question. Use an online cheatsheet. Note that the full mix will contain everything - let the overheads glue it together. Set each individual mic have an EQ that captures the essence of that drum:
    • Kick drum - low frequencies important here, but some highs to get drum character.
    • Tom drum2 - keep low frequencies and some mid.
    • Snare drum - keep mid frequencies and some highs.
    • Overheads - remove the low frequencies and most of the mids, keep only highs. The overhead mics glue the drum kit together through the middle - the low end is dealt with in their own tracks.
  • On the snare drum add the free VST "snare buzz". It is fantastic to bring out that snare sound.

Electric Piano

Since it is electric it's already perfect, leave it alone. The only thing you might want to do is compress it if you want to squish the dynamics, or adjust the EQ to create some space for another instrument.

Acoustic Guitar

Generally this is already coming in from a pedal, so like the piano you might not need to touch it. However I rarely find an acoustic guitar that has a frequency response I like, so I usually add EQ to remove boominess and boxiness. It's also a good idea to remove everything from around 80Hz and below, acoustic guitars can't hit those frequencies.

Vocals

Ah, the most challenging instrument of all. Compress - if there are really peaky parts, use two compressors, one to scrub the peaks, the other to smooth it out without crushing it.

I love a saturator on a vocal. Korneff Chocolate Milk (tape saturator) is great, GSatPlus (tube saturator) is free and also quite good.

Add in a further compressor to do sibilance de-essing. In ReaComp, in the detector, adjust the Lowpass and and Highpass so the detector is only working on a frequency range where the esses are, say around 7k-11k for a female. The attack should be fast and the release fast-ish, say 1ms and 30ms.

EQ - varies from person to person, but 120Hz for fullness, 240 for boxiness, 5k for presence, 8k for sibilance, 10k for air. Sweep the mid to find frequencies you like/dislike

Sax

I found that the sax is dynamically unbalanced even between notes. I compress it a lot. Either go for a little bit of reverb, or add some back slap delay if you don't want to muddy it with reverb. EQ this for sure.

Fine tuning and Mastering

Envelopes

Right click on a track header, click Envelopes and then choose say Volume or Pan. Now you can drag the fader/pan around as the song progresses. I normally do the 4 click trapezoid to bring a section up or down. You are almost re-producing the song with this - you can say drop a solo instrument right down so it isn't competing with the vocals, and then crank it up during insturmental sections.

Edit

Don't be afraid to split tracks up! You can realign missed tempo notes, or copy sections from another take or verse to punch out bad sections. Overlap all cuts so they seamlessly transition. Normally you can drag around the cross-fades, but if the fade isn't showing, right click the track, select Item Properties, and then type in new values for the fade in and fade out.

Mastering

The JS 1175 Compressor is nice to glue it all together if it feels a bit unstuck. Use the ReaXcomp multiband compressor to make it pop. Some very minor EQ to tweak up anything that presents itself in the final mix (because you were too keen EQ'ing individual insturment). Top it all off with the JS Event Horizon Limiter/Clipper, set the ceiling to 0, and wind down the threshold a little bit to make it louder but don't go too far or it will start to distort.

Handy Functionality

SWS/S&M

Action - SWS/S&M Copy FX chain from selected track (bind to key I)
Action - SWS/S&M Paste FX chain to selected tracks (bind to key O)

MIDI double hits

I found my older cheaper Yamaha DTXplorer was giving double/triple hits, probably picking up hit bouncing (they are pretty firm rubber pads) but I probably have the sensitivity and threshold too low in the drum head unit.

But all is not lost! I created a script that will remove hits that are clearly too short. The rule is if it finds a hit that happens too quickly after the same note hit then it removes it and stretches the first hit to be the same length as the two hits together.

In the MIDI editor, select all the notes. Under Actions, change the section to Main, and run a lua script. point it to my script and run it. Yah!

Channels

Generally, each track will only have 2 channels shown in the mixer. Even for mono, because FX will often create stereo. If however you have ended up with say 32 (and the track is really wide) you can fix it by clicking on the Route icon, set channels 2.

Timeline

If you want to make an arbitrary point your new zero, right click on the ruler -> change start time/measure - > Set to zero. You can later undo this with a Reset Project Start Time.

VST Drums

Steven Slate VST drums (SSDSampler5) is popular and a great solution if you don't have access to a real drum kit. If you don't have a midi input and just want to draw on the grid:

  • Add a track and add the fx SSDSampler5
  • In Create double click on a kit to select it
  • Click somewhere in the track Insert->New Midi item
  • Double click on the midi item to bring up the MIDI editor
  • Set the toolbar to note names
  • File->Note names->load note names and point to an SSD drums note name file

If you do have a drum kit and use midi triggering through an interface, don't forget to set Reaper enabled on MIDI inputs.

Sound Interface

If you are using an interface to record directly into Reaper, and you want to be listening as you go, then minimum latency is critical. I can hear a latency of 10ms and it is too distracting - you can't pluck guitar strings and have more than 10ms delay and remain sane and calm. You have two choices - have a setup with low latency, or don't listen to the signal, just record.

To get low latency (I'm in Microsoft Windows land) you can try using WASAPI, you might be fine, it's pretty good. You will need to tweak with buffers (keep them small, ie using less time, but requires a faster computer). What you really need is equipment that supports ASIO. Note that ASIO4all is not actually ASIO, it is a fudge, like WASAPI it might be okay. You need to specifically buy equipment with its own ASIO driver, such as the stuff from Focusrite, Presonus, various Behringer gear. Unfortunately at the moment it is not cheap.

Note: if you experience noise while connecting your laptop to your interface over USB, you might have a ground loop issue. If you unplug your charger and the noise goes away, then definitely a ground loop issue. Try a different charger, run on battery only, or put a galvanic isolator on your USB between your laptop and your interface to eliminate the gound loop.

JAW's Quick Notes To Himself for M32 mixes

  • Copy WAV to local drive, name the folder yyyymmdd <leader> - song 1 - song 2 - song 3 - song 4
  • Run ..\Rename x32 hex to dec.cmd
  • Drag WAV to new Reaper session, sequence time positions on a single track
  • Run JAW's Explode & Track Rules Parse.lua
  • Delete inactive tracks and rename tracks according to the prep standard
  • Using Ripple Edit mode, cut out bits not required
  • Save as yyyymmdd-1.rpp
  • Consolidate tracks into FLAC: Set output directory correctly, project name as yyyymmdd-2, set consolidated filename as $project-$track
  • Isolate vocals: after doing the consolidate there will be 4 flac tracks for the congregation in the root directory.  To pack and isolate them run
    ..\ProcessCongregationIsolation.cmd
  • Run JAW's Pre Mix Prep.lua
  • Optional: Phase alignment on drums (pull all tracks right a tad to allow space). Use a short section that features plenty of snare (bleeds into other mics), MAKE SURE it is "Adjust each item separately" and Adjust "Position". Use snare as the master.
  • Drag isolated vocals in and add the fx chain.
  • Double track acoustic guitar/etc if required.
  • Rough fader re-adjustment: GIVE YOURSELF HEADROOM!
  • Solo and tweak individual drums if necessary - levels, amount of gating, limits.
  • Going forward - don't be afraid to split tracks and do coarse volume adjustments on those (pre-fx) items.
  • Add in Bass and tweak.
  • Add in Piano, Guitar and any solo instruments and tweak.
  • Solo the vocalists, add in fx if not already there and tweak.
  • Make the START of the song really hit.
  • Run through the VSX checks.

  • Note: these are some of the decent isolators:
    Demucs:
    demucs -n htdemucs_ft "cong.wav" --two-stems=vocals

    Melband Roformer is possibly slightly better but takes ages:
    audio-separator "cong.wav" --model_filename "melband_roformer_instvox_duality_v2.ckpt" --mdxc_segment_size 256 --mdxc_overlap 4 --use_autocast --output_format=WAV

Sunday, 10 May 2020

More musings on Musescore - my Musescore tips

In my efforts to convert all my scores/tabs to MuseScore, I've got a few tricks that I better write down for you and for me so I don't forget. MuseScore makes my tabs look so pretty:


But they don't look like that out of the box, so here's a few things I do so that my tabs look this great!

Note: This was originally written for Musescore Version 3, I have updated for a few differences in Version 4.

Stave Properties

Right click in any bar and select Version 3 Stave Properties Version 4 Staff/Part Properties. I set up the instrument to be Acoustic Guitar (Tablature) and then tweak the defaults. If you need to go to Drop D or other tuning do that in Edit String Data, otherwise press Advanced Style Properties.


In Advanced Style Properties change the font to "MuseScore Tabs Sans" font size 10pt. Nobody ever complained for having a big clear font. However the spacing gets big so change the line distance to 1.75sp.


Now click on "Note Values".I set Font to "MuseScore Tab Modern" which gives you a plain "TAB" at the start. I then set the Stem style to Beside stave (which defaults to below) which is clearer than the tab numbers having stems. I set Minums "As short stem" as I find the double slash looks a bit cluttered. That's it for the Stave.


It's all XML!

A MuseScore file is of type .mscz which basically means "MuseScore Zipped", and inside the zip is a .mscx file which basically means "MuseScore XML", which is simply a text file that the program reads and edits...but being text, you can also read and edit it. Or hack it with tools and stuff. But generally XML implies good standards and techniques, one of them is the XML properties.

From File select Score Properties, I go for workTitle, arranger, composer, creationDate and copyright is my website. I also add a field called lastUpdate because I like to know when I created a tab, but also the last time I made some changes on it. File timestamps are not always meaningful.


What this gets you other than the files being index-able by "systems" is that you can have a page template that refers to these properties. So the header and footer of my tabs all have the following, repeated on each page.


Templates

XML naturally lends itself to templates. There are sooo many style options in MuseScore it is ridonculous. I had used MuseScore for a few years before I understood 10% of them, and I still only understand maybe 15% now. Go into Format->Style and edit away, once you are done you go Format->Save Style, and in the future you can apply your template to any other score with Format->Load Style. But this is only the styles - the previous discussions on Staves isn't part of a template, as with a few other things, I haven't quite got my head around how that works. Advanced Styles appear to have a template but there does not appear to be a built in editor, you may have to hack the XML. But don't, the easiest way to create a new score with the same normal/advances styles is to save-as, rename, delete all, and begin from there!

But back to template styles, here are some important ones, in the Format->Style dialog:
  1. Page - Music top margin, Music bottom margin. Similar to above, tweak to squeeze more out of a page.
  2. Version 3:Page - Min. and Max. system distance. This the distance between the first row of bars and the next row, that is, "system". I try to keep my tabs single page so I will tighten it if I'm chasing just one more system without spilling onto a second page. Tweak both these numbers and watch the results!
  3. Version 3: Bar - Spacing Version 4: Measures - Spacing ratio. Version 4.5: Spacing - Horizontal spacing - Spacing ratio and system density. These are great number to tweak to get your score onto one page, or if it's just two big to space it out nicely for two pages.
  4. Lyrics. Traditionally they go underneath, I have stuck with that. Lyrics are important because they help get the melody feel of the arrangement. Use Min. bottom margin and top margin to help with your spacing. Version 3: Use Text Style -> Lyrics Odd Lines/Even Lines Y spacing to get the tightness you want - also applies to Chords. Version 4: Format->Style->Lyrics and adjust Min margins and Min distance to cram the words up if you are aiming for one sheet.
  5. Additionally, in Format->Page Settings you can set Scaling Staff space to a lower number to really squeeze more out of a page.
And here's how I get my header and footer:




Other Good Stuff

  • Version 3: The best sound for play back is "Acoustic Bass", go to View->Mixer and in the Sound specify Acoustic Bass.
  • Version 4: The best sound for play back is a soundfont. Download a Soundfont for Guitar, this super simple nylon guitar soundfont is okay. Copy the .sfz or .sf2 file into the Musescore Soundfont folder, (is set in Edit->Preferences->General->Folders) and then copy the Soundfonts folder into a file explorer window. Restart Musescore, then open your sheet, View->Mixer, find your instrument, click Sound->Soundfonts and your soundfont will be there.
  • Version 3: Get a good Swing feel by dropping the text type "Swing". You can then adjust the sound with Properties->Swing Settings although the default of Quaver/60% is great. Version 4: has the same but in Format-Style-Score.
  • Add a Capo: 
    Version 3: place any text on the first note (eg "Capo at x") then go to Properties -> Capo Settings
    Version 4 (early): place text as per version 3 right click -> staff text properties -> Capo Settings and set the capo you are chasing. 
    Version 4 (later): In Palettes search for Capo and then place Guitar->Capo
  • Percussive notes, with an "x" head,
    Version 3: press shift-x,
    Version 4: click on the note, click on properties, change the notehead type to "x".
  • Version 4: Set Edit->Preferences->I/O->Buffer size to 4096 to get smoother sound playback.
  • Version 4: Turn of chord playback by clicking in the cog in the top right and untick "Play Chord Symbols". Note that this is an application setting, not a score setting.  To force the score to never play a chord you have to click on each chord symbol, properties, untick Play.
  • Version 4: the beam height of the notes underneath the stem are sometimes a bit long.  There doesn't seem to be a way of setting this in the style, but if you ctrl-A to select everything, then click on Note, then click on Beam, then click on Show More, you can then decrease the height of the beam at the start and finish.

That's it for now. I will have to pop up some more tips about using MuseScore, I had a bit of a learning curve from PowerTab, and to be honest, I'm still not that fluent. MuseScore for me is reverse Polish notation - I think of the note pitch first then the note duration rather that the other way around so I find myself pressing escape and "n", "q" or "w" a lot...