Looking forward to it, Normen! Thx.
D.
Celemony's new Melodyne 4 - holy crap!!!
- Soft Enerji
- Posts: 407
- Joined: 16 Jan 2015
- Location: East Lismore, NSW Australia
I use Editor........that's all I need really.
Tehe, okay so I imported the whole first half of our concert (56 min.) into Melodyne standalone, 8 tracks of drums, vox, 2x guitars and bass.. Melodyne calculated for a whole 9 hours on a 4 x 2.4GHz i5 with 16GB of RAM. Then I went in to see how it tracked the tempo of the single songs and I wasn't too surprised to find that it was completely wrong with the beats and tempi
Furthermore it doesn't handle such long projects very gracefully, I can't even play the project properly, at first it plays the tracks and then some tracks aren't audible anymore and for others it seems to play only partials of the sound.. Seems to be mainly a RAM / HDD load issue, the CPU isn't working too much. Well Celemony says that it does need a lot of RAM for long song files and even with lots of RAM it might not be able to handle it but I wanted to try it out anyway. Seems like the manual was right yet again
Pitch recognition was halfway okay but because of the bleed on all mics it has a lot of false positives, the automatic algorithm selection also didn't really work, basically all tracks have been recognized as "polyphonic", even the drum tracks. After setting the vocals to monophonic and the drums to "percussive" it had to calculate those tracks again (which worked quicker then though). The guitars and bass are DI so theres no bleed on them but the pitch recognition on the heavily distorted guitars isn't too great either.
All in all I kind of knew this wasn't going to work out perfectly with all the bleed, distortion and the long files but I thought I'd put Melodyne to the test here and it failed miserably as expected Next up I'll be exporting the single songs and see how it works then.
Furthermore it doesn't handle such long projects very gracefully, I can't even play the project properly, at first it plays the tracks and then some tracks aren't audible anymore and for others it seems to play only partials of the sound.. Seems to be mainly a RAM / HDD load issue, the CPU isn't working too much. Well Celemony says that it does need a lot of RAM for long song files and even with lots of RAM it might not be able to handle it but I wanted to try it out anyway. Seems like the manual was right yet again
Pitch recognition was halfway okay but because of the bleed on all mics it has a lot of false positives, the automatic algorithm selection also didn't really work, basically all tracks have been recognized as "polyphonic", even the drum tracks. After setting the vocals to monophonic and the drums to "percussive" it had to calculate those tracks again (which worked quicker then though). The guitars and bass are DI so theres no bleed on them but the pitch recognition on the heavily distorted guitars isn't too great either.
All in all I kind of knew this wasn't going to work out perfectly with all the bleed, distortion and the long files but I thought I'd put Melodyne to the test here and it failed miserably as expected Next up I'll be exporting the single songs and see how it works then.
So in regard to the project at hand I resorted to using Logics pitch correction (monophonic) on most of the vocal tracks, simply because Logic handles the whole concert (2h 20min) in one project without problems and editing is much easier than in Melodyne. I did compare the results of Logic vs Melodyne for two songs before switching completely and while Melodyne did handle the high end of the vocals slightly more transparently the results of Logic were at least in the same ballpark. For messups in the guitar and bass tracks we simply re-recorded these parts. Neither Logic nor Melodyne were able to properly recognize the pitch in the distorted guitars well enough to correct them.
For two songs where the vocals were particularly problematic I resorted to using Melodyne though. Interestingly for the most problematic parts Melodyne caused the same kind of artifacts albeit a bit less pronounced. As for beat detection - Melodyne handled recognizing the tempo of single songs quite well but in the end I realized I don't really need the tempo to mix the project so I skipped that completely.
The sound editor allowed some interesting tweaking but with the live tracks and the bleed the results were not really even enough to be useful, too many artifacts. I'd probably have to correct all recognized notes to remove false positives / overtones for it to work properly. And honestly, for a live recording it doesn't make much sense to go that deep anyway.
All in all Melodyne still is what it used to be, something to make good studio recordings perfect - not a wonder machine to mangle any kind of recording.
For two songs where the vocals were particularly problematic I resorted to using Melodyne though. Interestingly for the most problematic parts Melodyne caused the same kind of artifacts albeit a bit less pronounced. As for beat detection - Melodyne handled recognizing the tempo of single songs quite well but in the end I realized I don't really need the tempo to mix the project so I skipped that completely.
The sound editor allowed some interesting tweaking but with the live tracks and the bleed the results were not really even enough to be useful, too many artifacts. I'd probably have to correct all recognized notes to remove false positives / overtones for it to work properly. And honestly, for a live recording it doesn't make much sense to go that deep anyway.
All in all Melodyne still is what it used to be, something to make good studio recordings perfect - not a wonder machine to mangle any kind of recording.
- EnochLight
- Moderator
- Posts: 8407
- Joined: 17 Jan 2015
- Location: Imladris
This has been my experience with Melodyne as well, but to their defense - their guides always instruct to do the correction on dry signals with no effects processing. Obviously this is challenging if you're processing a live performance, but I thought it important to mention.normen wrote:Neither Logic nor Melodyne were able to properly recognize the pitch in the distorted guitars well enough to correct them..
Win 10 | Ableton Live 11 Suite | Reason 12 | i7 3770k @ 3.5 Ghz | 16 GB RAM | RME Babyface Pro | Akai MPC Live 2 & Akai Force | Roland System 8, MX1, TB3 | Dreadbox Typhon | Korg Minilogue XD
Sure, I get why thats the case, theres just too many overtones. The same is true for the AC/DC songs our singer sings with the Brian Johnson voice and while we *could* have recorded the guitars without effects (we both use Eleven Racks and no amps) thats not really possible for the vocals in that case Though the AC/DC songs are all pretty good, no need for correction there.EnochLight wrote:This has been my experience with Melodyne as well, but to their defense - their guides always instruct to do the correction on dry signals with no effects processing. Obviously this is challenging if you're processing a live performance, but I thought it important to mention.
-
- Information
-
Who is online
Users browsing this forum: No registered users and 12 guests