Jump to content

Latency Between Key Press And Note Sound

Recommended Posts

I wonder if anyone has a measure for the period of time it takes between a button being pressed and a note being sounded?


I realise that this will vary from concertina to concertina and even from reed to reed, but what I am trying to find out is what would be an acceptable maximum value for this latency on a good quality instrument.


I am asking because in building a midi concertina there are various places where delays occur and the sum of these delays should not exceed this maximum acceptable value.

Link to comment
Share on other sites

I had a very serious latency problem with the first version of my MIDI software, and found out, that above a certain amount it effectively cripples the ability to play, because sound is generated a bit later than you expect it and this confuses brain too much. From my experience, this critical delay is about as long as it takes your finger to fully press the button (when you have normal force acting on the bellows) - the sound should play no later than when button is fully depressed. Otherwise you have to take this latency into account when fingering (on my Elise this applies to lowest, weighted reeds which are significantly slower to speak than the rest) which requires "forward thinking" and not just following the melody.

Link to comment
Share on other sites

For my MIDI concertina (for which I use an iPad as synthesizer) I bought a JamBox Bluetooth speaker. However the bluetooth protocol introduced a latency of about 0.4 seconds, which was sufficient to wreck ones ability to play - hearing the previous note when your fingers had moved on! Using the wired connection to the speaker solved the problem (but reduced the freedom).

Link to comment
Share on other sites

Hi Don, the start transient of a free reed, as used in concertinas and accordions can in general be considered to be several tens of milliseconds.


In more detail, this transient involves at least two processes. One is to establish the wave form characteristic of the sound field, and the other is to establish the volume of the note. In my experience with accordion reeds, these two events can take considerably different time durations.


In my observations looking at microphone traces with audio software (such as Audacity), it seems that the reed tongue vibration must first establish its fundamental mode of oscillation, and the sound field with the pitch and timbre we recognize becomes established from that oscillation. Establishment of the steady state volume of sound can take considerably longer. There is often a volume overshoot, which then settles down to the steady state value. Also, the tongue vibration, vibrating as a cantilever, during this start transient, probably contains overtones, and there is also a possibility that a torsional mode is excited. I do believe that the start transient is an inherently unsteady process. As with other tone generators, it’s likely that this transient contains the shedding of eddies and vortices, and because of the turbulent nature of these processes, it’s not too surprising that these other modes of tongue vibration are excited. Incidentally, the overtones of a vibrating cantilever are not harmonically related (they are not multiples of two).


The following data that I obtained from accordion reeds may be useful, giving the note identification, the frequency, and the duration of the time necessary to establish the tone and timbre of the note, not the transient to establish a constant volume. The number in parentheses is for different reeds, but of the same pitch, when a tone chamber (cassoto) is used.


A2, 110 Hz, 97 ms (50 ms); C3, 131, 147 (46); A3, 220, 45 (60); C4, 255, 86 (50); A4, 442, 40 (32); C5, 512, 17 (36), A5, 878, 41 (25); C6, 1035, 61 (21); A6, 1768, 86 (82)


There is perhaps a trend here, where on average, the presence of a tone chamber causes a shorter start transient, concerning identity of the tone. The tone chamber also usually causes a larger initial sound amplitude, which then takes a while to die down to the steady state tone. For instance, for the C3 note, it takes an additional 168 ms for the volume to establish its steady state value. The tone chamber gives a more sudden sound, and it’s curious to me that concertinas do not contain such a chamber. I suppose it would make the instruments unacceptably large.


For the purposes you mentioned, I think you can consider the start transient to be something like 50 - 60 ms for the above frequency range. Reeds with lower frequencies than those will involve longer duration start transients. In fact, during the 20th century, the free reed was discontinued for decades as a sound source in pipe organs because of its relatively long start transient. However, later in the century, people brought it back, mainly because they liked its timbre, which contrasted with the sound produced by the other pipes’ edge tone generators.


But the start transient of the free reed is much more complicated than I can describe here. Cottingham did a study that some may be interested in (http://www.public.coe.edu/~jcotting/PMA035061.pdf ). In this study, done using free reeds from a Williams American reed organ, he found that the start transient of one reed lasted about 100 ms. He also found overtones of the cantilever transverse bending mode, the first torsional mode, and the first transverse bending mode that is perpendicular to the main bending mode (in the plane of the tongue!). He notes that the torsional mode of vibration is involved in the very start of tongue vibration and suggests that it might be able to exploit it - by tongue design - as a way to reduce the start transient. In a separate thread, a few of us discussed at length how a free reed starts vibration, though none of us considered the presence of more than one mode of vibration. Our little friend baffles us again. I should note that the typical reed tongue of a reed organ has a bend in it, near the tip, bringing the tip closer to the slot. This bend is absent in concertina and accordion reed tongues.


Best regards,



Link to comment
Share on other sites



Thanks for this. A lot more answer than I expected!


So latencies of a few tens of ms are probably OK? I plan to do some testing to time how much latency I can detect, but I would expect other people to detect different latencies.


I read that while really bad latency is unacceptable, such as in a BT connection, most musicians can adjust to a small-moderate latency that is consistent, but what they cannot handle is jitter - a variable unpredictable latency.



Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

  • Create New...