Recording an Accordion

[oldbayan]

I attach 2 of those little clip-on instrument mics with a mini goose neck, one for each side. This way I don't have to be stuck in front of a mic for the duration of the recording. I put the clips on the folds of the bellows, and can pick the ideal position for treble and bass sides.

 

[debra]

Based on Jerry's suggestion I made my latest recording (BWV1055R movement 3) using 96kHz 24 bit and I must say that the final mix down to 44.1kHz/16 bit sounds fine but not different from my other recordings. The input files are 3 times the size compared to starting from 44.1kHz/16 bit tracks, and I discovered some bugs in doing copy/paste within the n-Track mixing software I use, but other than that I found no difference. I guess that because I am mixing 5 (stereo) tracks down to one stereo CD track any perceptible difference in the individual tracks disappears.

 

[JerryPH]

Based on Jerry's suggestion I made my latest recording (BWV1055R movement 3) using 96kHz 24 bit and I must say that the final mix down to 44.1kHz/16 bit sounds fine but not different from my other recordings. The input files are 3 times the size compared to starting from 44.1kHz/16 bit tracks, and I discovered some bugs in doing copy/paste within the n-Track mixing software I use, but other than that I found no difference. I guess that because I am mixing 5 (stereo) tracks down to one stereo CD track any perceptible difference in the individual tracks disappears.

Well, using 24/96k or 24/192k or even 32-bit floating/48k files, these are all standards that far exceed human hearing in terms of specs and ability. Consider it like taking pictures with RAW data and then processing and exporting as JPGs. Prettty much EVERY camera on the market today has that ability and even most cellphone cameras now support the ability to take very high resolution or RAW format files. I've been playing with nothing but RAW format files for as long as it's been out on all my Nikon cameras (since the 1990's). The differences are indisputable.

As far as bugs in copying pasting... I will politely suggest to look at upgrading your software, not all software is equal. Reaper is something that I've used for years and gives me professional results, and has a workflow that not even the tens of thousands of dollars of expensive software like ProTools (the professional recording standard in all high end studios today) offers (loading speed, ease of use and ability to accommodate the needs of the users instead of forcing the user to work in it's more limited workflow).

Also, a "trick" is to not export down all the way to the "bottom" of the range, but record, play and output at higher levels. For example, YouTube accepts 24-bit/96k files in their recordings... and thats what I will put in to those files (I started doing that late 2024, I like what I hear!). Though one has to be super careful with output levels, else YouTube screws with your audio and removes a lot of your hard work making it sound like a low quality audio file.

Moving down to 44.1k quality files lowers a lot of the dynamic range. Consider it akin to saving a picture that goes from white on the left to black on the right in 4 steps versus 100 steps... the one with 100 steps is going to be smoother and less evident gradations in between the steps.

You cannot realistically post process files that have low dynamic range or data in it without loosing more quality when you export to it's final format, but it is easy and better to use the largest files possible to retain all the nuances and then export them to a lower quality format. I find that it is FAR easiest to post process without introducing additional noise (mostly heard as hiss). EQing also becomes a lot easier and more pleasing to my ears using higher quality files than lower quality files. Ultimately, saving them in the lowest possible quality levels WILL strip out some of that quality, thats normal and should be expected.

Of course, your equipment has to support them and you need speakers/monitors that can relay this information to you accurately, else you won't hear it very well if at all.

Audio post processing is no harder or easier than post processing pictures or video, if using the highest quality possible, its easy to go "less" but if you start from the bottom, its impossible to up from there.

Not that I have the hearing of a dog or bat (lol) but I can clearly hear the difference between my accordion recordings of the same clips that are 44.1k and 96k files... there is a smoothness and life that is lost at the lower quality recordings that isn't there at 44.1k recordings. I attribute that to the greater dynamic range. I can also do things like remove furnace noise and background air movement and my voice won't sound anywhere near as robotic, something that does happen when using lower quality files.

Here is a small video based on the OPINIONS of someone else, but they have some nice facts in there too!:

 

[John M]

Well, using 24/96k or 24/192k or even 32-bit floating/48k files, these are all standards that far exceed human hearing in terms of specs and ability. Consider it like taking pictures with RAW data and then processing and exporting as JPGs. Prettty much EVERY camera on the market today has that ability and even most cellphone cameras now support the ability to take very high resolution or RAW format files. I've been playing with nothing but RAW format files for as long as it's been out on all my Nikon cameras (since the 1990's). The differences are indisputable.

As far as bugs in copying pasting... I will politely suggest to look at upgrading your software, not all software is equal. Reaper is something that I've used for years and gives me professional results, and has a workflow that not even the tens of thousands of dollars of expensive software like ProTools (the professional recording standard in all high end studios today) offers (loading speed, ease of use and ability to accommodate the needs of the users instead of forcing the user to work in it's more limited workflow).

Also, a "trick" is to not export down all the way to the "bottom" of the range, but record, play and output at higher levels. For example, YouTube accepts 24-bit/96k files in their recordings... and thats what I will put in to those files (I started doing that late 2024, I like what I hear!). Though one has to be super careful with output levels, else YouTube screws with your audio and removes a lot of your hard work making it sound like a low quality audio file.

Moving down to 44.1k quality files lowers a lot of the dynamic range. Consider it akin to saving a picture that goes from white on the left to black on the right in 4 steps versus 100 steps... the one with 100 steps is going to be smoother and less evident gradations in between the steps.

You cannot realistically post process files that have low dynamic range or data in it without loosing more quality when you export to it's final format, but it is easy and better to use the largest files possible to retain all the nuances and then export them to a lower quality format. I find that it is FAR easiest to post process without introducing additional noise (mostly heard as hiss). EQing also becomes a lot easier and more pleasing to my ears using higher quality files than lower quality files. Ultimately, saving them in the lowest possible quality levels WILL strip out some of that quality, thats normal and should be expected.

Of course, your equipment has to support them and you need speakers/monitors that can relay this information to you accurately, else you won't hear it very well if at all.

Audio post processing is no harder or easier than post processing pictures or video, if using the highest quality possible, its easy to go "less" but if you start from the bottom, its impossible to up from there.

Not that I have the hearing of a dog or bat (lol) but I can clearly hear the difference between my accordion recordings of the same clips that are 44.1k and 96k files... there is a smoothness and life that is lost at the lower quality recordings that isn't there at 44.1k recordings. I attribute that to the greater dynamic range. I can also do things like remove furnace noise and background air movement and my voice won't sound anywhere near as robotic, something that does happen when using lower quality files.

Here is a small video based on the OPINIONS of someone else, but they have some nice facts in there too!:

That Video was informative for me Jerry. Thanks!
He sure likes 46K sampling over 96K -- At first he said some thing like Drums, Orchestral & blue grass was better at 96K but then said 48K was the best. In general 96K was too perfect, "clinical", than 48K.

 

[JerryPH]

Yes, for “boom-crash” kinda music, he said he liked a lower resolution… logical if you think about it, the kind of music he records is inherently loud and very dynamic, he hears it as to “clinical”. I would classify it as being more “authentic”. As I said… different people have different opinions, so to each their own.

He did say to just avoid 44.1 in all cases. I disagree with that. If you like your sound after testing higher resolutions, stick to what you like. The one thing that I know about me, I would never sacrifice sound quality for size, not only will I use higher resolutions in my videos, I use .WAV files, which are lossless, unlike MP3 which are further compressed (.flac are also ok but not as well supported), and are huge, but size is irrelevant in today’s world of high speed internet for my needs/use case.

Only exception for me would be live streaming, then I could see dropping to 44.1k or 48k, because of streaming provider’s restrictions, and in those cases it’s less critical to have maximum sound quality.

 

[dak]

Yes, for “boom-crash” kinda music, he said he liked a lower resolution… logical if you think about it, the kind of music he records is inherently loud and very dynamic, he hears it as to “clinical”. I would classify it as being more “authentic”. As I said… different people have different opinions, so to each their own.

He did say to just avoid 44.1 in all cases. I disagree with that. If you like your sound after testing higher resolutions, stick to what you like. The one thing that I know about me, I would never sacrifice sound quality for size, not only will I use higher resolutions in my videos, I use .WAV files, which are lossless, unlike MP3 which are further compressed (.flac are also ok but not as well supported), and are huge, but size is irrelevant in today’s world of high speed internet for my needs/use case.

Only exception for me would be live streaming, then I could see dropping to 44.1k or 48k, because of streaming provider’s restrictions, and in those cases it’s less critical to have maximum sound quality.

I tend to work with FLAC files which are also lossless but compressed. They tend to take up less than half the size of WAV files. For sending recordings around, I tend to use WEBM format (often without a video track) and the (lossy) Opus codec at 176 bps (which tends to work pretty well for accordion). Lots of smartphones and computers can play those fake videos even though they barf at playing "proper" native Opus files. They have the codec but only accept video containers for it (WEBM uses the "Matroska" container format while native Opus files use an Ogg container that is more often associated with the Vorbis audio codec).

My main workflow is 96kHz/24bit. Opus uses perceptual compression and as one part of perception it says "nobody hears stuff above 24kHz anyway" and samples down to 48kHz. AV1/Opus (for video/audio) is what Youtube is working with usually.

 

[JerryPH]

I attach 2 of those little clip-on instrument mics with a mini goose neck, one for each side. This way I don't have to be stuck in front of a mic for the duration of the recording. I put the clips on the folds of the bellows, and can pick the ideal position for treble and bass sides.

I think that's great if it works for you. Depending on the quality of the mics, they can either sound really good or fall victim to the "so close that I cannot evenly catch the entire range of the keyboard. I found that using mics on floor stands strategically placed are not visible in the field of vision of the accordion being recorded and I can get up and turn off/on the camera without needing to "unplug". There are good and bad points to be made on all systems.

 

[PeterFromAkita]

When you say lifeless, that might be the reason I'm asking in the first place, just something felt like it might be missing. My ukulele only recordings feel the opposite to me but I'm guessing it's because the accordion may need different treatment due to being a very different instrument. Any ideas on how to change that?

Interesting discussion.
In order to make a room more lively, you basically have three options.
1. Record somewhere else
2. Use reflection.
3. Use less reflection and use artificial means to liven up your recording.

For using reflection, you can buy diffusers. I use diffusers from Hofa. They are not cheap though, and you need quite a lot.

For using less reflection you can buy absorbers. See the same brand. The key to absorbing reflections is that you want to absorb reflections over bass, mid and treble frequencies alike. Treble is quite easy, but for lower frequencies you need mass. I use bass traps by the same brand. A cheaper alternative would be to stand a thick mattress in the corner at a 45 degree angle. Not ideal, but it might work. Then in post processing your audio you can dial in small amounts of reverb. Mind that you take a short reverb that suits your recording location. Especially when making a video it would be weird to have a small room sound like a cathedral. I used to record a lot at locations that are not ideal. I made use of the 8 pattern of the ribbon mics (point one end to the musician or ensamble, and the other to the far away ceiling) and then added stereo artificial early reflections to the recordings. Again, price is an issue, we used Bricasti for this, but it can increasingly and tastefully be done with the effects in your computer multitrack system.

 

[JerryPH]

When people market expensive things, often what happens is that someone comes out with a cheap way to do the same thing for a lot less but it doesn't look as pretty.

In that vein, a simple wood frame off of 2"X2" or 1"X3", often on caster wheels to make it moveable and cheap moving blankets are the most effective poor man's way of countering reverberence in a room. I've gone to Harbor Freight and bought about 15 cheap moving blankets as protective covers for my winter-stored car. You can look in thrift stores and get the big heavy thick ones for 10% of their retail costs if you are lucky.

Just for testing purposes, I hung a couple up in my living room made from the frame of 4 light stands and 2 cross bars. It does make an audible difference, but the closer, the better. You kind of need to box yourself in to this area for best results. For about $75 you could make two rollers out of wood and cover a fair sized recording area.

If my basement was not as nicely controlled as it was, I'd not hesitate to add a couple of these blankets in a setup for recording.

Another small thing that I learned but did not really cover in my video is that even in my area upstairs where I made with the video, the closer the mics are, the less echo they picked up and of course a more obvious one is to test different angles as some parts of the room are much more prone to echo than others. My angle for this video is about the best possible for this setup. If I had moved to the couch, the echo would have been way more audible with the mics that were further back and on the camera mic.