5

I use GarageBand to write music but in theory this question could apply to any composition and recording workflow.

A problem I often run into is that different audio players seem to have their own "flavor" of output. For example, whatever I write obviously sounds like what I intended in GarageBand. When I export it to MP3 format and listen in QuickTime Player, same result.

When I import the same file in iTunes, it seems to apply some sort of treble booster that makes the whole track sound tinny. Ditto on my iPhone (where the default music player is basically a subset of iTunes). In practice, I "fix" this with a bass booster or treble reducer in the Music -> Equalizer menu. I'd expect the default media players on Windows, Android, and Linux to have their own specific quirks.

My current approach is to test on a small selection of hardware and software that I have at home, although obviously it's impossible for me to test on every possible piece of equipment my music could potentially run on.

It gets even more difficult if I have to account for different types of headphones and audio equipment. Some headphones and speakers are well-balanced, some might be bass-boosted, and others might have low-quality tinny audio. The amount and type of padding between the ears and speaker holes on a pair of headphones will obviously cause some sort of frequency-based distortion or bias. I'd like to be able to hear bass frequencies on my music loud and clear on my headphones without throwing an equalizer into the mix. I'd also like other people, some of whom might use heavy bass-enhanced or noise-canceling headphones, to be able to listen to the same music without voiding the warranty on their eardrums in the 20-100Hz range. It's not as simple as "make all the bass instruments louder".

Different headphones have different audio leakage profiles as well - for example, the ones I use leak a lot of noise through from the outside. I consider this to be a feature rather than a disadvantage since it keeps me aware of my surroundings. Meanwhile, someone else might be using a noise-cancelling headset that blocks everything except the music.

On top of that, I'd like to build my music so it can be reasonably listened to in highly variable environments. Indoor or outdoor, and also on mass transit with a lot of possible engine noise polluting the lower audible frequency spectrum. Do NOT assume the listener is using a noise-canceler. My experience so far is that a bass booster or treble reducer is a must for listening on a noisy train, for example. Any way to fix this without causing the reverse problem when there isn't a noisy train engine humming loudly in the background?

The main question: Is there an industry-standard way to calculate frequency distribution and instrument volumes for the optimized audio quality across a wide range of hardware, software, and environments? Preferably a workflow built for a DAW but anything reasonable should be adaptable.

EDIT: Writing music is a hobby for me. My production budget is only the time I spend on it, the cost of my computer and associated hardware (to be fair, I use that computer for a ton of other stuff; GarageBand comes free with a Mac), and the electricity it runs on. I definitely don't have the budget allocation to bring a recording engineer on board, except as a volunteer.

Improve this question
5
  • 1
    Interesting question! Looking forward to see the answers! Maybe you'll get more stuff on sound design…
    – Tom
    Commented Aug 17, 2020 at 19:21
  • I didn't know that was a thing on Stack Exchange. Did it just pop up recently?
    – user1258361
    Commented Aug 17, 2020 at 19:28
  • 1
    That's another good question!
    – Tom
    Commented Aug 17, 2020 at 19:34
  • 3
    You're talking about mixing and mastering, both of which are non-trivial arts really. :) It's HARD and it takes SKILL and EXPERIENCE. :D I'd be greatly surprised if someone could tell you how to do all that in a Stack Exchange answer, but let's see!
    – piiperi Reinstate Monica
    Commented Aug 17, 2020 at 20:05
  • 1
    @user1258361: "I didn't know that was a thing on Stack Exchange. Did it just pop up recently?" – The Sound Design Stack Exchange started in the Spring of 2010.
    – Jörg W Mittag
    Commented Aug 18, 2020 at 13:31

5 Answers 5

Reset to default
3

iTunes has a feature called "Sound Check" which scans all newly imported music and does some auto-EQ and volume adjustment so all the songs in your library play at roughly the same volume and have similar EQ profiles. This may be your issue. If you don't want it to do that, it's a check box in the app preferences.

Beyond that, I agree with the other answers regarding a good studio monitor pair or other high quality flat response speakers to ensure that what you hear matches what you're recording.

You may also want to check your iTunes preferences to make sure that it is not automatically converting imported audio to a default lower quality MP3 to save space.

Improve this answer
2
  • I already use Sound Check for exporting my projects. I've also tested using varying quality of MP3 output and I hardly notice any difference except if I'm using the 64 kbps encoding. A bit of advice: the 64 kbps encoding produces horrid results, only reason to use it is when you need to save every last bit of disk space.
    – user1258361
    Commented Aug 18, 2020 at 16:49
  • This is definitely the problem for the itunes case: you are allowing itunes to "remaster" the track. The tinny nature is almost assuredly a bitrate and quality change (for the worse) where the high frequencies start to "sparkle." It is editing the files.
    – Yorik
    Commented Sep 18, 2020 at 14:32
2

The main question: Is there an industry-standard way to calculate frequency distribution and instrument volumes for the optimized audio quality across a wide range of hardware, software, and environments?

Yes: hire capable and experienced mixing and mastering engineers. That’s the industry standard.

How do those engineers ensure mixes will translate well? Mainly, they know their business. They have worked for years developing their ears and their knowledge of how a mix should sound if it’s going to translate well. And they also know how to test their mixes and masters in their custom listening environments and a limited number of other environments to correlate the data gathered by their ears and make the necessary adjustments.

So far, no algorithm, software, hardware, or simple process has been invented that can replace the work of good engineers.

As in pretty much every area of human endeavor, the hobbyist should not expect to operate at the same level as the professional. Having been a semi-pro/low level pro engineer for several years in the past, I would expect to take ten years or so of daily part time recording, mixing, and mastering to reach the level I was at in my prime, when I was able to get hired locally to record and mix. If you primarily prefer to be a musician and/or composer, I suggest that you focus your time on that and do your best on the engineering side while looking for people who either value your music highly enough to pay for professional engineering or who are engineers themselves and are happy to work with you on spec.

Improve this answer
6
  • I tend to disagree with your assessment that "the algorithm" can't replace mixing and mastering engineers; the only thing preventing that is a lack of research and application. I'm sure "the algorithm" could produce much better (and automated) mixing if Big Tech productively dumped a few billion bucks into the research effort.
    – user1258361
    Commented Aug 17, 2020 at 20:29
  • 1
    @user1258361 I would tend to agree with except that, IMO, mixing and mastering are part of the art of music making and, in certain bands the sound engineer, plays a significant role in the final production, which would be different with another SE. An algorithm might product good mixing, but everything might end up too "standard"…
    – Tom
    Commented Aug 18, 2020 at 9:11
  • 2
    @user1258361, you are misquoting Todd's answer. He stated that so far none has been invented. Secondly, there does exist a "standard" way to cal f-dist, and dB blah... That uses the GOF-FFT. But a human in the loop needs to make a choice as to what is important in that spectrum. Perhaps using samples of known sounds or physics based models could be used with AI (e.g. ANN) to make the choices but this has not occurred yet. And this would need training which requires a human to make a choice.
    – user50691
    Commented Aug 18, 2020 at 13:23
  • @user1258361, All that is needed to screw that up is for different groups of people to make different choices about what is an important signal feature.
    – user50691
    Commented Aug 18, 2020 at 13:23
  • 1
    Kind of an echo or expansion on what I think @ggcg is getting at: mixing and mastering are (partly to mostly) artistic processes, by which I mean trying to remove the human element would not lead to the most pleasing results. They are not mechanical process (despite the use of the word “engineering” which I think was an unfortunate choice made long ago). Just as we don’t currently have algorithms or mechanical devices that can write songs or play emotional guitar solos, for the same reasons we don’t have algorithms or mechanical devices that can produce effective mixes or masters.
    – Todd Wilcox
    Commented Aug 18, 2020 at 15:30
2
  1. If you export sound from a program and it sounds clearly differently in another program then there's something wrong. Perhaps you have an equalizer running all the time in one of your programs? Of course there is some flavour added by the MP3 format itself, but at high bitrates it should be rather small. If you observe a big effect, then there is an issue elsewhere. Normally it shouldn't happen.
  2. An "industry standard" for listening in audio production is called studio monitors. These are speakers intending to have flat frequency response and reproduce sound neutrally, without coloring the sound, as consumer equipment typically does. But to make it work you need you need to make your software work properly as well (point 1.)
  3. It is a good exercise to check how your mix will sound on various consumer type of equipment, headphones, laptop speakers, mono radio, car speakers – but this can only supplement listening on good monitors.
Improve this answer
1
  • I'm almost convinced that different MP3 player apps on various platforms include their own flavor of subtle post-processing. That's why I test on a small selection of platforms that I have at home for purposes of "cross-platform compatibility".
    – user1258361
    Commented Aug 18, 2020 at 16:52
2

Know your equipment.

Buy a decent set of reference monitors, and learn how they sound in relation to other environments and equipment. Use them consistently and regularly, so you are aware of how they will work regarding the music you record.

Remember that people who like a lot of bass will already have the bass eq turned up to silly levels on their equipment, so you don't need to worry about that. If it sounds good on reference monitors, in-ear headphones and a car stereo, you're as close as you'll get without paying an experienced mix engineer to redo everything.

Improve this answer
1

Apparently, AI-powered auto-mixing has been a thing for a few years now. As you probably expected, it's a tradeoff. If you "outsource to the robot", you save money while giving up customization and flexibility you would otherwise get with a professional recording engineer. I don't expect "the robot" to displace professional recording engineers out of their jobs anytime soon, but auto-mixing definitely provides a "much better than nothing" option for independent musicians on an extremely low budget.

https://www.theverge.com/2019/1/30/18201163/ai-mastering-engineers-algorithm-replace-human-music-production

Landr is one of the most popular such services, hosted as a web service. You can upload the song you want mastered, let Landr’s algorithm analyze it, choose between 3 options for how strongly you want effects applied, and then export the result. It’s a catchall approach, and it’s not exactly flexible.

Software company iZotope approached AI with an educational lens. The company already makes a popular suite of plug-ins called Ozone, and added in an intelligent “Master Assistant” in 2017. The assistant doesn’t do all the work for you. Instead, it gives you a starting point you can tweak. That way, producers can make informed decisions based upon the choices the AI has made.

Improve this answer

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.