I've long been impressed with the progress that MuseScore is making and this seems like a huge jump forward.
> Under the design direction of Martin Keary, we are making significant improvements to the interaction models and interface of MuseScore 4
For those interested, Martin Keary (aka Tantacrul on YouTube) creates these very impressive YouTube videos of reviews of music notation software. His reviews are usually very scathing, but an honest critique.
After his video reviewing MuseScore, the MuseScore team spent a lot of time addressing the issues he pointed out and eventually hired him!
I just watched the MuseScore video. I get that he’s trying to be entertaining, but he just comes off as a jerk to me. Doesn’t make him less correct of course, but do we really need a blood-vomiting Shostakovich or imagined email exchange making fun of “New All” (among other examples)?
Honestly, it’s disappointing that this is the default way to criticize someone’s work, however bad it may be.
Imagine if your users, teachers or even parents critiqued your work in this way.
I'm willing to give it a pass because it's in good faith and he gives praise as well; it's not like an Angry Video Game Nerd-style rant. Also, this is a review for his YouTube audience more than it is criticism aimed at MuseScore's designers. I'm sure he wouldn't critique the work of his kids or students this way.
Tangentially related but super cool, Tantacrul was part of a project[0] with David Bruce where 5 composers of various backgrounds were given the same brief (basic themes, instrumentation, etc.) to see how different people came up with different ideas. Each of the composers are also worth their own follows as well.
This idea was taken from Andrew Huang (who himself took it from a photography channel) who puts the same sample in front of different electronic music producers[1]
The original "4 Photographers Shoot the Same Model" [0] series by Jessica Kobeissi that Andrew Huang's series was inspired by was nominated for a Shorty Award [1] in 2019.
As soon as I saw this pop up on HN, I thought “hey, MuseScore — isn’t that the notation software Tantacrul loves to hate?”. I’m not a composer, and I’d never heard of let alone used MuseScore before stumbling on his videos a couple of weeks ago. What a great surprise to see they’ve been listening to their critics!
After finding his channel I slowly worked my way through watching all his videos. One of my favourites is about “sonification”, or creating sound from of data. Might be quite interesting to the HN crowd: https://youtu.be/Ocq3NeudsVk
Even in his video he said he liked many parts of MuseScore — he never said he totally hated it. His critique of it was much more entertaining than scathing.
His opinions and tastes are somewhat odd imho, so despite some good observations, his opinions ought to be taken with a grain of salt. For instance, in his Dorico video, he doesn't seem to appreciate that some of the software's design stems from a separation of content and appearance (much like TeX), which has has long term benefits to the user that may not be apparent at first.
Although it may not really fit what you're looking for, there was a one-hour video linked here a few days ago that critiziced the entire modern desktop GUI landscape.
After I watched his videos I thought the same thing. Almost every big piece of software would benefit greatly from a Tantacrul-like teardown, but I also realize that such a thing would require a whole heap of real-world domain knowledge and possibly hundreds of hours of use to track down all the issues so thoroughly.
Only tangentially related, but EEV Blog recently did a video on PCB design BOM (Bill of Materials) optimization where he took a random open-source project as a counter-example.
I'm not an electrical engineer, nor do I design PCBs, but it was still pretty interesting if you're into that sort of critique.
I love Tantacrul's videos. The humor is great in general, and when you deal with terrible UIs on a daily basis (like I am on my job), it hits close to home.
They really are great. As much interest as there is in user interface design, and as much as it dominates our lives as designed artifacts, the field has very few serious critics. Tantacrul’s videos are a great template of how to make interaction design criticism both thorough and riveting.
Exciting news! This feels like a Blender 2.8-style project.
What's funny about this is that a composer/music YouTuber/UX designer called Tantacrul meticulously roasted MuseScore's UI on YouTube, and the MuseScore team ended up hiring him, leading to this project.
This UX review video is awesome and hilarious, and also illustrate the competitive advantage of combining general design expertise with specific domain experience.
As an occasional user of MuseScore (which I like a lot), I'd be weary of them going down the DAW path. They really should focus on music notation.
There are already more DAWs than needed, some free, some inexpensive, and most of them pretty good.
It's unlikely a MuseScore DAW will bring anything new to the scene, whereas on the music notation front they occupy a unique and very strong position (their competitors being either outrageously expensive or not good).
It doesn't need to be "a DAW", it just needs to be "the parts of a DAW that lets you make the sheet music sound good". If they pull that off, which to date literally no one has, then I couldn't care less if it wasn't suitable for general music production without the sheet music side in the slightest. Need to be try to be FL Studio, or Ableton, or Logic Pro, or Cubase, or Avid Pro, or Studio One, or Reaper, or... Do what they don't, because they're not going to.
If MuseScore will let me use Spitfire's BBC Symphony Orchestra and gives me DAW-style control over how my sheet music renders to audio with the finesse that score alone doesn't allow for, then that would be something we've never been able to do, and it's about time someone did it.
Genuinely curious, why is that better than just having good midi i/o? I would think almost anyone doing composition and notation, who has expensive, realistic orchestral libraries to play things back on, already has a DAW of some type. And the type of full control you're describing sounds like it could really include a lot. You'd need features like a recording timeline and playback control, parameter automation, advanced routing options, return tracks, side chain inputs, plugin latency compensation, ability to freeze or record plugin output to save cpu resources. Pretty much everything a daw has.
I've long been a fan of PreSonus's Notion (and I say that as someone who has a lot of experience with Sibelius, Finale, and Dorico). It's one of the few products out there that pays attention to the "parts of a DAW that let you make sheet music sound good," and I (personally) find the UI easier to use than most. I won't say it's the best thing out there if you need beautifully engraved parts for a large orchestral work and you need super high levels of organization, but I love it for composing.
I do agree, though, that no scoring program really let's you have the level of control over the way sheet music is rendered at the level that Spitfire or VSL allow. There's always some kind of disconnect, and I always end up tinkering with things in a full-fledge DAW in the end.
I don't think they will be going for 'full DAW' functionality, rather taking the same path that Dorico [1] has taken - in allowing DAW-like features to produce better audio output from the scoring application, while still remaining in that same app.
I've found that Dorico has allowed me to control things that otherwise I would have needed to put into a DAW to do (such as alter played note lengths to my satisfaction while retaining the notation I think is appropriate) - something which was previously difficult or impossible to do.
In addition, being able to have DAW-like control of a mix removes the need to transfer the project (or stems) to a DAW, which can be great, particularly when you want to quickly create something for someone's assessment; yes, in the final version you may well need everything that a DAW offers, but often you'll just need a bit of control to spruce things up a bit and take it in the right direction, and having greater control over this (via VST Instruments and hopefully effects) will allow this all to happen in-app rather than having the (sometimes long-winded) transfer to a DAW. Which can then need to be done again when there's a significant enough change to the music to mean a re-import.
A DAW that focuses on notation rather than piano-roll doesn’t really exist in the market however. The UX of the apps that support a stave is stuck in the 90s.
How does OpusModus compare? I'm less than amateur when it comes to music and found it from completely different direction, so I would like to know opinions.
One plus of Reaper is it runs on the same platforms as MuseScore. Everything I've heard about Logic Pro is great, and I would like to try it, but it's Mac-only.
I tried reaper for a day and could not figure out how to do anything. Switched right back to live.
I think it's a promising project but he urgently needs to hire a UX expert.
+1 for platform independence though. I can't understand how logic users lock themselves into the apple ecosystem only to complain about the expensive RAM upgrades to keep Reaktor snappy.
I agree. I have recently started taking composition lessons for large bands (e.g. 18 pieces), and I've been using MuseScore as my main tool.
Seeing it go the way of a DAW is disappointing, because there's a fundamental difference between composition and producing - that is, the difference between an abstraction of ideas (a lead sheet) and a final product (a render).
Getting either of these to a competitive level is extraordinarily difficult, and getting more than one is almost impossible ; I'm willing to be proven wrong, but I think this change of focus may end up hurting the project.
TL;DR: It's not about having another DAW. This functionality is also important for workflows that have nothing to do with producing.
Yes, some of that new stuff looks a lot like a DAW, and for DAW users the first priority should be to integrate the notation software with existing DAWs. However, this is different: when you don't write notation for the sake of controlling virtual instruments, but for performers on live instruments, it's still very useful to be able to ‘preview’ scores and parts thereof through soft-synths. You don't need production-quality sounds, but you do need a faithful reproduction of the notation — something that understands ‘pizzicato’ etc. To that end, software such as Sibelius has suitable synths already integrated, which can often spare you the hassle of cobbling the scoring program together with a DAW.
This is also relevant for collaboration. A couple of years ago I gave MuseScore another go because I wanted to facilitate collaborating with a Linux user. The program itself looked promising enough, but I also needed playback functionality (neither of us can play all the required instruments) and getting something cross-platform to work was such a nightmare I gave up after a day of fiddling with soft synths. It had gotten to the point where I was running Linux in a VM and tried to hook together the Jack in the VM with the Jack on the host system. I found software that was able to do this, but only for MIDI, and another program that could do it, but only for audio. It was way too janky. We resorted to just synching MusicXML and doing playback completely separately.
(Meanwhile, there now is a Linux version of Reaper on the horizon, but no idea if its notation component has gotten any good yet. Anyway, it's neither free nor Free, so the idea of telling everyone in a band/ensemble to get that isn't as attractive.)
From that perspective, the potential of MuseScore 4 is obvious. I'll be keeping an eye on it.
Having used both MuseScore and Reaper in my workflow, I initially agree that scoring and DAWs are two different things, and that I could simply export MIDI files from the notation software then take let the DAW do the rest.
But I see the point of how good playback can affect the composition process, something notation software is not known for. Although I believe instrument quality should not be a cover up for bad compositions, there are undeniable instances where otherwise good compositions sound bland or uninspiring when played by the "lightweight" sounds bundled with notation software (as they are not DAWs). At times, after hearing it sound right within the DAW, with my favorite plugins and sound libraries, I am inspired to tweak the composition a bit, via the piano roll.
In such scenario, if playback could be improved, it would be nice if a DAW can be removed from the compose-listen-tweak loop. At the very least, have realistic instruments and realistic (humanized) playing, even before moving work into the DAW.
Seems ironic, didn't MuseScore start as a sequencer called MusE that had some minimal notation abilities, which were separated out to create MuseScore?
Open source music notation is a thankless job - notation is hard, and when you're composing, you run into corner cases pretty quickly.
I'm personally a big fan of lilypond because I'm also a programmer (and I also like LaTeX, which has a similar model), and I love that with lilypond I get both superior output, and source files that I can track with git, which is great when I have different versions of some of my works - I can actually compile them.
Browsing the git repo above shows how scores and parts can be structured - from the same codebase, that repo can generate the parts for the individual orchestral instruments, similar to how many other notation packages can, except with lilypond I have total control. And it's the only notation solution I've tried that makes me comfortable that I'll have decades of compatibility - I can't even open most of my old Finale/Sibelius scores from college since I stopped paying for upgrades. Many composers also believe that Lilypond has the best engraving quality of all notation software.
But lilypond's been stuck on MusicXML support for years, and with pretty poor sound output options. On the other hand, sound output isn't a major priority for me, I usually use a sequencer or a piano to sketch out ideas, and I think there's something of an impedance mismatch between notation and sequencing anyway.
For MuseScore to go in this direction implies that they must be thinking they're already nailing the notation side. I'll try it out at some point to see if the UI ease-of-use makes up for the other things it loses - maybe I can use it for lead sheets - but I kind of doubt they're moving in a direction to be able to support serious composers.
> For MuseScore to go in this direction implies that they must be thinking they're already nailing the notation side.
They're not. There's still a lot of work to do. It's passable, but it's not 100% yet.
When you're scoring with Lilypond, what front end do you use ? And does that frontend have midi in / out ? I don't really care what the internal sounds are, but when I'm scoring/writing I like to be at my piano, and being able to play back chord voicings/rhythms is helpful.
I personally use Frescobaldi - it works well for my flow. Apparently it has some midi input support, but I haven't tried that myself. Midi output/generation has also existed in lilypond for a while.
I think that sounds a bit like asking for interoperation between Latex and Microsoft Word: technically kind of possible, but only in limited unidirectional ways that won't get much use in practice.
Exporting a limited/ pared down version (say just the "content" minus formatting) to a tidy and comprehensible (human readable/ editable) .ly file would be a great feature. Probably even feasible for a casual dev to write such an importer/exporter as a plugin. Might look into it.
I’ve lost a lot of faith in
MuseScore over the last couple of years. Their UX still has some major warts over Finale that haven’t improved, and they completely screwed over their users by encouraging them to upload their compositions to MuseScore’s cloud library and then cutting off free access to the cloud library. They ginned up as much content as they could from users and then announced this huge change in access a few months later.
I’ve lost a lot of faith in this business model because early users get screwed the hardest and as time goes by, it’s harder and harder to justify using this app. Why is it that these projects so often promise the world early on only to continuously chip away at their own promises and undermine the faith of the community that gave them business to begin with?
> Why is it that these projects so often promise the world early on only to continuously chip away at their own promises and undermine the faith of the community that gave them business to begin with?
As they say, never attribute to malice that which is explainable by incompetence.
I've seen this kind of thing 100's of times when some naive people get into product development. The team promises life-time subscriptions or other completely unsustainable business models because it _feels_ like the right kind of payment plan. They genuinely want to give their users the best features at the lowest prices.
Then the reality of maintaining and updating a large software product/service sets it. Engineers are expensive. Marketing is expensive. Operations are expensive. What started off as wide-eyed "wow, we're making a million dollars!" turns into "oh no, we only made a million dollars". Expenses start to rise faster than income.
Then someone who actually knows how to run a business steps in and does the Picard facepalm while reviewing the companies pricing plans. How did they ever expect to build a business with such a totally unrealistic model?
So the now battle scarred founders are forced to face the music. Change their pricing and product offering or watch their company evaporate. They aren't doing it to be mean or to squeeze blood from a stone, they are trying to keep their business solvent.
I expect that navigating this transition is one of the hardest things a fledgling business will ever have to do.
They did. OP also neglected to mention that this was forced by the music industry threatening to sue them into oblivion if they didn't put it behind a pay wall.
Musescore did have separate listings for PD or CC-licensed compositions, vs. "no copyright is intended, wink wink" ones. They could have allowed access to the ones that were properly licensed.
Yeah. Lots of things that were uploaded to this cloud service were transcriptions or complete rips of sheet music of in-copyright works. Sheet music is also copyrighted and when the music industry found out about the service, it got strong armed.
There's a lot of sheet music that's absolutely in the public domain, though. Most of what's on IMSLP falls under that. (IMSLP only provides raw scans however. There is an Open Scores project that's supposed to work on making machine-readable versions of those in Musescore, but I don't know if their output is freely accessible outside their cloud platform.)
If you listen to a MIDI video game track and use your ear to write a violin-and-piano arrangement, do you own the copyright to your arrangement, or does the original composer?
The original composer still owns the copyright for the composition but you own the copyright for the performance. Just like if you sang someone else's song.
Much like cleanroom-written code, these are often made without ever looking at the official sheet music, just hearing some synth midi and figuring out a neat way to replicate the general idea using traditional instruments.
But yes, apparently it would be violating US copyright law to rearrange music this way.
There's no way that were completely unaware of this possibility when they started the service. They did this intentionally to get as big as they possibly could before the music industry came after them, leaving their users high and dry in the process.
You can't start a music platform without having some sort of peripheral awareness of how litigious that industry is. They could've tried to build it out more responsibly early on and chose not to.
The initial idea of the freely available library was wrong in the first place. You can't legally offer copyrighted stuff for free download. (or maybe you could if you paid all the copyright for each download)
I don’t think Musescore falls in the “move fast” category at all. They were simply naive (and underestimating how much pop music would be transcribed and uploaded). In any case, I found their communication very clear.
Musescore has significant UX issues and things have been getting worse with each release. Instead of adding new complex functions that are already done better by others they should try to fix the basic things first.
What are some of the UX problems you're thinking of? Personally I've been pretty happy with it so far. (My main point of comparison is Sibelius, which I used until I went all-Linux three or four years ago.)
Oh there are some. Some of them have been fixed in MS3
The one that's almost a show-stopper to me and I realized it only recently: if you insert notes (between existing notes), it won't move the existing notes to the right, but it will instead create this "long" bar with more notes than your time signature.
There are so many weird things like that, but the one that infuriates me the most is the inability to move notes left and right. The answer is to cut and paste, which - if you have already written out multiple bars but realize one note is in the wrong place or you’re a quarter note off - becomes an exercise in frustration.
I recommended this software to someone, and my first words were: “This is a powerful program, but it has got a steep learning curve. It feels like it was designed and written by a programmer who also plays music”
The answer from this person two weeks later: “You were totally right. Very frustrating.”
Scoring directly in the software is a huge pain + buggy midi support often forces me to use a different software and then import the midi into musescore for printing.
I'm still on 2 too, mostly because I recall being unable to use the plugin interface ionsome early version of 3, in late 2018; not sure if it was being reworkwd on or whatnot, but since I had been using MS to author some books and lessons for my students, I was already relying on the scripting, with no time to learn something new or wait for the APIs to work again. Since I'm stuck in this version limbo, I use a library I wrote in Python to manipulate mscx files. I think thos is also a mini-rant on how little FOSS software seems to care about backwards conpatibility. Still, v2 still is much better than any other notation software I tried so I'm not switching.
I find this move a bit weird to be honest. It seems like they are targeting the wrong users. People interested in writing sheet music usually do not need a DAW. And if they do, they probably already have one.
Also, instead of focusing on polishing their product core (writing sheet music) they are diluting their efforts.
They are probably seeing the writing on the wall.... people who use computers to compose music (which is probably 90% of musicians under 40) don't tend to use sheet music except as an output format, and even that is declining along with the need for real musicians playing instruments.
Pianoroll notation and all the things that tend to come with that just lend themselves to computers.
In many ways this is like CAD transitioning away from fractional notation to decimal. Fractions suck when using anything measured or derived, as they give you a tradeoff between accuracy and readability, that isn't the case with decimals. (i.e. is 19/64 bigger or smaller than 9/32? Meanwhile it's quite obvious that .296 is bigger than .281) With fractions, and traditional music notation, quantization is an important and annoying part of the process.
This actually represents a major and fundamental philosophical shift in the sense that it takes Musescore from being a tool of representation to a tool of creation. I may be overinterpreting this of course, but I am reading a lot into how this might be following (or bucking) a general trend in internet/tech/wider culture that could be characterised as favouring mediums and platforms of consumption over ones of curation… yet alone creation.
Does MuseScore have feature parity with Finale wrt notation? Essentially:
Speedy Entry or its equivalent (keypad-entry ftw)
An escape hatch for the user to paint their way out of all western notational edge cases (splayed stems, chord clusters, crazy slurs, etc.)
If it doesn't have the first then you lose all the copyists (gotta go fast).
If you don't have the second then you don't get the contractors engraving Henle editions (because control over the score isn't fine-grained enough).
Edit: I initially wrote "reliable escape hatch" but removed reliable-- at least when I was using it, many of Finale's workaround tools were buggy. Stuff like dragging a shape and the shape drifts away from the pointer the further you drag it from the origin. But you could power through the bugs and eventually get exactly what you want and print it out.
What about the escape hatch for things not listed there? For example, I didn't see clusters, or [one of probably 20 other things I could list one of which almost certainly requires an escape hatch]
Clusters are easily created as a custom notehead. This gives you full control over exact design and placement.
As for the rest, well, without knowing what your items are, I can't really respond - but Dorico is AMAZINGLY capable.
Irrational time signatures? Any crazy/arbitrary/arbitrarily nested tuplets you care to dream up? True open/unmearsured time signatures? Arbitrary microtonal systems? Multiple simultaneous time signatures? All out of the box.
But of course it supports the ultimate escape hatch... you can import SVG or export the entire score as SVG.
Oh, okay, it imports SVGs just as graphics elements? I was jumping to the conclusion it meant interpreting as notes that become an editable part of the score.
For those that are not afraid of using a DSL to generate scores, Lilypond is great.
Much like using graphviz to generate workflows, you can just hop in an editor and start typing out your score, creating reusable chunks along the way. There is also a decent live-reload style editor called Frescobaldi.
I assume you mean Lilypond. It's great, but the interop with MusicXML format (the closest thing to a standard when it comes to machine-readable sheet music - and also supported by MuseScore) could use some improvement.
Holy moly! As a big user of midi (I'm trying to make music for embedded platforms) the sequencer is a KILLER feature.
I've used LMMS in the past as I still find it much easier to compose with a sequencer than an actual music sheet but the major flaw is it's midi export not exporting the instruments meaning I have to touch it up in Aria Maestosa.
Ps. The musescore soundfont is probably the highest quality open source one out there, not a single instrument sounds wrong. It's my goto soundfont for composing
Oh man, I could have really used the sequencer when I took my music composition classes in college. Finale and Sibelius are not conducive to actually writing music.
The thing the most frustrates me about Musescore is the core dev-teams irrational refusal to add scroll-bars to the score panel (even as an option). Requests in the past have always been met with a “you are doing it wrong” response https://musescore.org/en/node/268908
Has new input on the design changed this approach?
Want to add that they really need to revamp their note entry process, which is very unpleasant and inefficient, and necessarily involves a lot of backtracking and redundancy. I hope this is included in the next version.
I've used MuseScore on and off, mostly because it's UI frustrates me and gets in the way. It's powerful, but it needs a way to make simple notation easy. If they're truly going to make it better for casual users, I'm all for it.
Exciting times. I appreciate the team's work on creating such high quality software. It doesn't feel like some random open source software. But it's always a work in progress, at the same time.
Every application is a work in progress. They either improve continuously or in point releases, step by step.
We were using musescore for our sheet music while I was playing in a symphony orchestra and, it was very capable back then. It was missing some very niche things (like Turkish Music Coma notation which divides a note to 9!) but, it was vastly usable.
I'm glad they didn't stop and moving forward. It's a very nice piece of software.
If you already play an instrument or sing, learning the fundamentals of music notation will take less than an hour, and practice makes it fluid. Following scores along to music or repeatedly puzzling out how to play pieces from the notation is what makes it automatic, and happens over time, just like any other skill. You start kind of laboriously with counting lines to see what note it is, and quickly become able to recognise notes directly.
Most notation programs can import MIDI, including MuseScore.
> Under the design direction of Martin Keary, we are making significant improvements to the interaction models and interface of MuseScore 4
For those interested, Martin Keary (aka Tantacrul on YouTube) creates these very impressive YouTube videos of reviews of music notation software. His reviews are usually very scathing, but an honest critique.
After his video reviewing MuseScore, the MuseScore team spent a lot of time addressing the issues he pointed out and eventually hired him!