Audio detection is maybe one of the more popular (or I should say ‘frequent’) queries on the board. Pending getting info on your o/s and DAW as @ed66 says, a search on the board about this will produce a plethora of posts. However, they may not give a solution, but they possibly might add insight. The process has some quirks (or maybe it should be ‘nuances’.)
@ed66 were you really posting at 06:45 ?? My alarm clock is still in deep zzzzz at that time
OS is Win 10. I have 3 DAWs, all producing the same results. I work mostly in Mixbus32C but tried the audio detection also in Ardour and Cubase 12 LE. All 3 got the similar results.
I detected using the audio file provided directly. I.e. - not live audio coming from a track. The chord results do not, to me, sound anything like the audio file when I play it.
I get similar results detecting from playing a track
The chords detected, as well as the suggested scales, don’t match the diagrams on the page. If they are supposed to then it didn’t work for me. If they created diagrams using different audio then it’s caused confusion but not wrong.
I’m in Linux at the moment but if someone enrolled in the class with access to the audio wants to compare chords I’ll drop into Windows in a little while and post those also.
It’s a synth chord loop. It’s bright, so it has harmonics which are likely to confuse the chord detection in Scalar 2 but that’s not my question. The real question was if the values detected using the provided loop were supposed to match the diagrams. That’s not a question someone who isn’t taking the class can answer.
I have Transcribe! so I can get a very accurate view what’s probably being played, notes and individual note volumes which gives you a view into how the performer most likely hit the keys. I’m using Transcribe! to help me build confidence in what Scaler is telling me, most especially when Scaler’s results don’t sound right.
I’m sorry to take up space in this forum. I think this would be better discussed in a class only forum with the instructors and other’s that have access to the same files. And please don’t get me wrong. I think the class is of a lot of value. I’m learning about Scaler features I didn’t know where even in there, much less know how to use.
I’ve posted several times on audio detection here, and Scaler operates in a different way to other apps in this area. It’s internal structure means that there are somethings it doesn’t do (by design, effectively) and some things that it does do much better than many alternatives
I posted a couple of links to some of my earlier postings here
I use it all the time for pulling progressions out of multi-bar synth sequences in DUNE, Omnisphere and others, and my workflow is described in the link above. As you see, my take is that it’s better at picking up higher chord extensions than most, but its lack of time awareness and way of dealing with single notes means more work than others. (BTW, the latter two are not bugs - it detects changes in chord sequences)
Thanks for your response and I strongly agree with the overall statements. Scalar lacks, IMO, much ability to communicate the time sense of what’s happening with chords, and as such intermingles melody, grace notes and mistakes with chords which can give some strange results. You might say Sclar is doing the right thing when it tells me F#maj7-F#maj9-F#maj7 but when you look at it in the time domain it’s clear that the F#maj9 was never intentionally played but rather was either a mistake or created by some synth harmonic for a 64th note. Interesting, but gets in the way (IMO again) of quickly choosing the right chords out of Scaler.
I don’t know DeCoda but will check it out. In the case of Transcribe! it’s a little time consuming to use but it’s able to really accurately show you the notes that are actually sounding over short periods of time in both the frequency domain as well as the time domain. I find it useful for detecting not only the chord but also left & right hands which can be very useful if someone is playing a key split, but also eliminating notes that don’t really contribute to the real chord being played.
Hi @LGTrader The course is run by the School of Synthesis so any questions regarding the course itself (payment, registration, login etc) should be directed to the School of Synthesis. Any questions about technical support or scaler functions and features can be raised here (as mentioned in the Scaler 2 Course registration email) because as you see we have a wonderful and active community who can be very helpful! Of course there is also Plugin Boutique whom are responsible for selling Scaler itself and they can help too!
If you set the threshold to the right position, as it was in the Scaler 2 Module 2 creative summary video, you should get the same results as in the video. While it might not be 100% accurate as to what was originally played in the loop, it should be a good approximation to get you up and harmonizing with the loop as Davide does in his creative summary.
Audio detection is something that we are working on improving and it should be used as guide rather than gospel at this stage.