Hey everyone. I’m fairly new to DJ-ing in general, so forgive me if I’m asking a painfully rookie question. I’ve got djay Pro AI on my Macbook Pro and using a DDJ-200 controller. I’m getting the hang of isolating vocals using neuralmix, yet they sound so tinny with harsh s’s. I do my best to use the EQ settings to reduce that and it’s still not great.
I have not yet ventured into using a DAW, so if that’s the answer, I would love any advice. Maybe I have to isolate vocals outside of djay, clean them up, and re-import them as their own audio files?
Thanks in advance for any guidance. My googling and searching gave me so many different videos to watch, it was overwhelming, so I appreciate some personal advice here.
I haven’t really compared how different tools isolate the vocals, but as you noticed that the result might not be really good.
The best option is to try to find the official acappella version from that song, but it so rare that artists release them. Some acappella versions might be available from dj pools and different online music shops.
let me just make sure I’m clear on what you want to do: You want JUST the vocal to be extracted from a song? to lift an isolated vocal track from a mix?
I’ll have more questions if I’m correct…but I can try to help.
Hey Janan, a lot of this will also depend on the quality of the original song. For example, you’ll get a much better result from a WAV or 320kbps mp3 that’s direct from an authorized digital seller. Whereas a YouTube rip or lower quality file (say 128kbps mp3) will give you a lower result from Neural mix. I hope this helps.
@dj_romy_fi Thank you! Yes, definitely vocal stems make life better and I’ll keep searching to find more and more. Cheers.
@heysoundude yes, for the sake of this question. And eventually, I’d love to be able to pull out lead riffs or anything melodic from a certain to mix into another. Though yeah just for this topic, I’m narrowing it down to vocals. Appreciate you.
@TheCunningOne Good point! I did not even think of that. I’ll go by and buy or download higher quality tracks and play around with that.
You’re welcome, happy to help and add ideas.
The suggestion to start with the highest quality/resolution tracks is the proper starting point, but then it gets to your use of Neural Mix. I would think that trying to get the isolation you’re probably looking for in one pass might be contributing to any sound quality degradation, but you DO have to realize there was some processing on the original vocal in THAT mix- if you’re looking for “chest tone” or “body” of the vocalist and that wasn’t present in the original, Neural mix won’t help to bring that back to compensate for any “tinniness”…so try pulling everything around the vocal down 6-9-12dB (tough without markings on faders) and printing THAT as a mix (WAV, not AAC) using the mix record function, then re-import and repeat as necessary, until you’re getting close to what you can work with.
(I recommend the feature to record your mix set in djay without ever having tried it - I don’t know if it works. If It doesn’t, I have a work-around. But start there.)
This technology is still new and doesn’t yet work how we as users would want it to most of the time. Currently, some songs work better and some songs less so and the things you can do to influence this are limited. Currently, DJ software tries to one up each other by making neural-mix-like features better and better but all of them still has some way to go before they sound “natural” all of the time. Here’s some more background to this discussion: What do you think on the new Stem separation in Serato
Neural Mix vocals usually sound better if you keep something else in the mix. If you Solo the vocals, keep the Harmonics or Bass or Drums active - it will sound more natural to the ear.
I wanted to thank you all who contributed to such great responses for @Janan, and for welcoming them into our community in such a positive way!
These are all great comments and I hope they were able to help point them in the right direction regarding their inquiry.
Thank you again and welcome again Janan!
Got it, thanks! I’ll go over and read that.
Yes, thanks for validating that method, as that’s what I was doing and wondering if there was something I wasn’t quite doing right.
I do that regularly too. Just a tiny bit of one of the other ‘stems’ makes the isolated part a lot better.
when you say “tiny bit”, can you give a rough estimation of how much below the vocal fader the ‘other’ Neural Mix fader is? 20dB? more? less?
It depends, I always do it by ear…
It does not have to be harmonics or drums from the same track, during mixing leave them sufficiently open of either track. Initially, I was a bit disappointed about the sound quality of isolated stems, but in practice just very useable indeed this way.
I saw someone say that this new tech is all bad because all it does is killing dynamic range. But in that regard, what is the good old cross fader doing?
One question, is the neural isolation from Djay Pro AI and the separate Neural Mix program the same?
If there’s an acapella that I must have, I’ve been using www.fadr.com to get vox, stems, etc
It’s only $10 a month/unlimited tracks/unlimitied storage and the results vary from track to track, but I’ve taken acapella and ran it through fadr.com a second time to remove artifacts.
If you don’t want to pay for fadr, there’s another tool you can download