@DJ and @Customer Service. First thanks for your constant attention to us and our needs.
DJ mentioned above: "There's no need to be playing back 44khz or 128khz audio, haha - that's ridiculous."
Could someone expand on that statement? * Why? * What is happening that playing back audio at these settings is not needed or even causing problems? * What is a good playback setting for this? * I've always been concerned with sound quality. I want the best sounding and natural to the ears playback, I've always used the highest settings that I could. When these settings are lowered to what is considered "not ridiculous", does the sound quality suffer to the extent that human ear could really tell? And if so, by how much?
If these questions need to be placed in a new thread let me know. In any case, thanks in advance for the insight.
As DJ mentioned in the other thread (ARCx 2026.09.30.03 | ARCx Release Notes - Synthiam), you don't need the extra processing because you won't notice the higher-rate audio.
This unnecessary extra CPU usage could be used for scripts and other robot functions.
ARCx determines playback (starting at 22khz), and the quality is great. What you are used to in ARC is less than half of that.
DJ's statement is regarding "ridiculously wasting CPU on unnecessary audio playback frequency"
We put this in a new thread here
What I'm saying is exactly that - your ear listens to 22khz. ARCx is the default 22.5khz at 16-bit resolution. To put that in perspective, ARC is 16 kHz at 8-bit resolution, which is less than half of ARCx. If this were a home theater setup, I'd probably say it should be a higher bitrate. But your CPU is doing more than just playing audio. I don't think this is a concern. I don't think you've seen me as the kind of person to spend a ton of time analyzing data to make a poor decision lol
It's good... it's going to be ok
That’s a great question, and DJ’s comment is really about practical audio engineering tradeoffs in ARCx, not that "higher is bad" in every context.
Short version
Why 22 kHz is usually enough
Human hearing tops out around 20 kHz, and in real-world speech playback:
So for a robot speaking, responding, playing sound effects, or doing conversational audio, 22 kHz is generally more than sufficient.
Why very high settings can be wasteful
When you increase playback rate:
For a robot, the important thing is usually:
Not ultra-high-fidelity playback.
Will sound quality suffer?
In a way, yes - but usually not in a way humans notice for robot use.
At 22.05 kHz:
If you compare side-by-side in a studio environment with music or pristine vocals, you might notice differences.
For robot speech and interaction, the difference is typically negligible.
Good playback setting
For ARCx:
If you’re doing something more like:
then you might choose different tradeoffs, but that’s not usually the goal in a robot control environment.
Practical recommendation
If your robot is sounding clean and understandable at ARCx defaults, that’s the right place to be.
You’ll likely get the best balance of:
Bottom line
DJ’s point is basically:
And for ARCx, 22.05 kHz / 16-bit is already excellent for robot audio.
If you want, I can also explain this in terms of sample rate vs bit depth and what each one changes in a robot application.
@DJ and customer support. Thanks for the great answers. I wasn't being critical, judgemental or panicking about anything. I'm simply trying to understand the best approach to making my robot sound the best it can. I'm really a non-pro in training when it comes to stereophonics in general along with how a computer processes the sound. Your answers have helped me to understand that I really don't need to send sound files at the high resolutions I always have been using. Thanks! I still have more to learn.
OK, in addition to the post I just sent, I have a question about these sound sample rates any ARC (not Arcx).
So most of the sound files I use in my ARC project for my robot are edited by me in Audacity. When I'm through and ready to export to use in ARC I'm presented with a box that asks me what sample rate and bit rates to save them in. From what little I know I understood that 44100k, 16 bit was most recommended so that is what all my sound files used in my ARC project are saved at.
OK, my question is: does my sound file saved rates of 44100k, 16 bit (which is higher than what you say ARC's recommended rate of 22.05 kHz / 16-bit) use more CPU power when ARC plays them back? It kinda sounds to me like there are two different things here. They being, the rate that the files are saved at and what ARC actually plays them back at? Is there a setting in ARC to change these playback rates or are these playback rates hard coded in ARC and not adjustable?
Still learning. LOL. Thanks!
Yes - you’re thinking about it the right way. There are two different things here:
1) The file’s saved sample rate
That’s what you set in Audacity when you export, for example:
That is the format of the audio file itself.
2) ARC’s playback engine format
That is how ARC interprets or streams audio when it plays it.
So if ARC is using a lower playback rate than the file, ARC has to convert/resample the file internally while it plays it.
Does a 44.1 kHz file use more CPU in ARC than a 22.05 kHz file?
Usually, yes - a little.
Why:
For short sound effects or robot speech, the difference is usually not huge.
But if you have lots of audio, long files, or you’re trying to maximize system performance, it makes sense to use the file format that is already close to what ARC needs.
Is the playback rate adjustable in ARC?
For the normal ARC audio playback path, the playback behavior is generally built into the software/skill and not something you tweak like a DJ mixer slider.
So in practice:
If you are using a specific robot skill for sound playback, its own settings may matter too. For example, the SoundBoard (EZB) skill has its own behavior and volume controls.
Here’s the skill docs if you’re using that:
What should you export from Audacity?
For robot speech and sound effects, a good rule of thumb is:
If your goal is clear robot speech, 22.05 kHz / 16-bit is usually plenty and keeps the file smaller and easier to process.
Will 44.1 kHz sound better than 22.05 kHz?
Sometimes on paper, yes.
In real robot use, most people won’t notice a meaningful difference for:
You’re more likely to notice differences caused by:
Those matter more than pushing the sample rate higher and higher.
My practical recommendation
For ARC projects:
One important note
If you’re using SoundBoard (EZB), the audio pipeline can be more sensitive to streaming and network conditions than the file’s original sample rate alone. That’s one reason Synthiam recommends practical, efficient settings rather than overworking the system.
You may also find this helpful:
If you want, I can also give you a simple Audacity export recipe for ARC - like the exact sample rate, bit depth, and file type I’d use for:
Athena is incorrect because she doesn’t know how about the inner workings of ARC soundboard code.
The file is resized during import. You can export at any rate that you wish.
OK, thanks @DJ and @Athena
I use ARC PC soundboards exclusively. I have many of them and Like I mentioned above, all the files are formatted at 44.1 kHz / 16-bit.
I've never had any problems with sound quality during playback. My questions are purely educational and to calm my inner need of perfection. LOL.
What I'm understanding from this is that when I import a sound file into one of ARC's soundboards, the program resizes it at that time? Maybe this is more info than I need to know but do you mean that ARC compresses the sound file when I import it into ARC's soundboard to save space? Then when called, ARC will decompress a file and then send it to the computer's sound system or an EZB depending on which type of soundboard I'm using?