Glad it was helpful . Your mileage might vary, but IMHO I don’t see anything to be “concerned” about…unless you are using Scaler for your music final.
I had planned a longer post on this topic a while ago because I use detection all the time and had learned a few things along the way. I see audio detection as what is… an inexact process which is highly dependent on initial state. I’ve never been concerned about variability of results and I certainly don’t see them as bugs.
Like any detector trying to tease information from a system not designed to report it, there will certainly be a range of results. Those ranges often introduce some interesting variation or musical paths that might not be apparent…one users bug is another users undocumented feature. And besides, at least from my perspective, those variations align with Scaler’s design intention of nudging vs. directing.
As you implied, forums like this one help tease out and then disseminate information, and helping set reasonable expectations is one of it’s key value props. So therefore, characterizing detection limitations as concerning or buggy might work against the goal of helping people explore the tool within it’s design constraints…just a thought.
These are great conversations and as you’ve probably seen the dev team is always eager to take feedback and cycle it into the mix. I’d wager that refine detection algorithms is a standing item on their scrum board.
Thanks @Ed1 Please don’t worry too much about my ‘concern’ it’s OK. Whilst I accept the reality of bugs though I am hoping it doesn’t mean I cannot flag things even if others like yourself are happy and/or content.
We disagree here. Feeding in an isolated piano loop and getting back incorrect or inconsistent musical data may not be what some would call a feature, documented or otherwise
I respectfully disagree. Limitations ≠ flaws, I can though grasp the concept of limitations.
And yes you’re right, @Ed1 from was incredibly gracious in stepping in and did so without any sense of reticence in dealing with my query even if it was user error.
I’d pretty much like to leave it here. The community is mostly great, but to flag again I am fine with having an issue and even happier now that the team are helping to figure out what’s happened.
Scaler’s audio detection does two things, frustrates me and surprises me. Sadly that’s the nature of the technology at a relatively inexpensive level, to go to Melodyne style detection you pay $500 but that is JUST for the detection. Scaler can only expect to do so much at $50.
Having said all of that, whenever I need to know which direction I am going in (scale, chord type etc) Scaler helps me 100% of the time and for this reason it is advertised as a bona fide feature.
It is not entirely useful but my detection video (if a little old) shows my genuine expectation and result.
Thanks for taking time to input! (I had actually seen this video already, helpful).
Whilst I don’t though have the 100% success rate you mentioned, I think I can empathize with your comment re “frustration and surprises” a bit. I respectfully disagree that “for this reason it is advertised as a bona fide feature” but anyway as mentioned I am quite happy to wait for @Ed1 to DM me back with any advice he has, even if user error.