Strange modulation on sounds

bearhack

Disciple of Aoede
Supporter
Defender of Defoko
Hello Everybody! I'm in the process of making my own VCCV voice bank using cz's tutorials, and I have all of my samples recorded and mastered, so now I'm in the process of setting up my oto ini files. I just wanted to test the first sample (-b@_b@b_k@b) to see if I did it right because for some reason the waveform doesn't look clear enough even in the spectrum for me to distinguish where the consonants are at times, so I just popped the first sample with a sample oto that I set up with the tutorial into utau and tried to render just "-b@". I get this really strange modulation in both the vowel and the consonant like I'm talking into a vocoder or something when it renders, and I don't have that problem with other voicebanks that I've used before. So, I went on to oto the next sample, (-b3_b3b_k3b) and tried to recreate that in utau - it sounded better, probably because the consonants were more distinct, but it still has that weird modulation (I did set the modulation down to 0 manually like cz said in the video too). Any thoughts on what might cause this? I'm wondering if I need to re-record all of my samples again because maybe I was too close to the mic, or singing too deep (I recorded the samples at F3) or maybe with too much resonance in my voice. It didn't seem that way in the samples though, and I was able to master them pretty well without too much trouble (all they really needed was a noise removal, eq, and to normalize the volume between the samples). But that was weird too, no matter how soft I de-amped the samples, it still sounded like I was shouting into the microphone unless I turned down the intensity to 40.
Any thoughts on what might cause my voice bank rendering to sound like garbage like this?
 

na4a4a

Outwardly Opinionated and Harshly Critical
Supporter
Defender of Defoko
can you upload an example of a sample that is strange?
If your consonants were not clear enough then it would still render, just with a very mild consonants sound so that wouldn't really cause other issues.

It sounds like you may of clipped you recordings but I wouldn't be too sure of that unless I got to see some of the samples.
 

bearhack

Disciple of Aoede
Supporter
Defender of Defoko
Thread starter
http://www.mediafire.com/listen/ani31e3h2dbed4h/-b3_b3b_k3b.wav
http://www.mediafire.com/listen/k9h4926b36h9okx/-b@_b@b_k@b.wav

Here are the first two samples that I just mentioned I otoed and tested in utau. You should be able to download them from my mediafire account since I don't see any place to upload them here. These are the raw versions of the samples, btw, just as they were recorded without any mastering. It doesn't look to me like the waveforms clipped when I looked at them in audacity. Jeremy, any help you can give me is greatly appreciated. Thank you!
 

Halo

Icon by Wanpuccino @ DA
Administrator
Defender of Defoko
Any chance you could screenshot what this looks like in your UTAU oto editor? Because I downloaded these samples and they work fine for me, so it's likely a technical issue on your end (and not a problem with your files).
 

bearhack

Disciple of Aoede
Supporter
Defender of Defoko
Thread starter
Sorry, perhaps I'm misunderstanding you Halo, but give me a moment...
I can't seem to embed the images for whatever reason. I'll just post them here:

Here's how the oto looks in setparam.exe:
http://www.mediafire.com/view/v4xlxzkvvp9r0z6/testoto.jpg

Here's what I did in utau to test it:
http://www.mediafire.com/view/9tu5tpe4ego4t43/testsample2.jpg

Here's what it sounds like:
http://www.mediafire.com/listen/0pd6bgo77gb7hco/testsample.wav

Any thoughts?
[doublepost=1450666103][/doublepost]http://www.mediafire.com/view/5mzt8u8hbehktwe/oto.ini
http://www.mediafire.com/download/63q8zctvt88txgy/testsample.ust

Pish, I'll just upload my oto.ini and .ust file too - maybe I'm usting it wrong, or maybe the oto is set up wrong
 

Zoku

making doper vocaloid music than the rest
Defender of Defoko
Hello! I can't test anything out right now since I'm out of the house and on mobile--but have you considered the resampler you're using? Certain voicebanks sound absolutely horrible on the default resampler but sound fine on their suggested resampler.
 

Halo

Icon by Wanpuccino @ DA
Administrator
Defender of Defoko
Okay, in setparam that looks normal.
Your UST is fine. And your output is average (maybe mine just haphazardly generated slightly better FRQ files). Now that I understand the sound you're complaining about, it... actually is probably your recording, maybe the resampler. But this is a relatively minor problem, not "garbage" as you said originally. The original file is kinda clipping a little (so Jeremy was spot on). Out of curiosity more than anything else, what mic are you using?
 

na4a4a

Outwardly Opinionated and Harshly Critical
Supporter
Defender of Defoko
Yeah that output is generally normal for a male voice.
You could look into using something other than your current resampler which all give various results. Any samples that are clipping should be recorded again but otherwise may be all set with a different sampler
 
Last edited:

bearhack

Disciple of Aoede
Supporter
Defender of Defoko
Thread starter
Did you see the waveform clipping? That's really strange because I was monitoring the gain while I was recording and didn't notice any clipping and I don't see that characteristic "squarish" on the peaks in the waveform in audacity when I look at it.
Anyway... I just purchased a new MXL USB 006 condenser microphone and pop shield to record with, I turned the volume attenuator way down and lowered the gain so that it wouldn't clip... Perhaps I am just being too picky? I tried using TIPS instead of resampler and it seem to come out slightly better, but it still has that weird modulation. Maybe I'm just expecting too much.
[doublepost=1450668441][/doublepost]Anyhow, I won't worry about it too much for now... Maybe it won't be so bad for the other samples and I can put it together into my new song. Thanks for your help, Halo, Jeremy and Zoku!
 

na4a4a

Outwardly Opinionated and Harshly Critical
Supporter
Defender of Defoko
How close do you record into your mic, if I may ask?
Where you record can also play a role in how good the result is.

From what I understand, since you said that you had to lower the gain even with the back switch set to low. You could possibly be recording too close to the microphone and a little bit of a strange tone can be heard in your recordings that might suggest this? (I am only assuming since I don't have any experience with the mic you own.)
 
Last edited:

bearhack

Disciple of Aoede
Supporter
Defender of Defoko
Thread starter
Well, in this case, I converted my 5x5 cedar closet in my bedroom into a miniature sound booth and set my computer up in there to record with my new mic because I really wanted a dry environment. I have to admit that it was kind of... cramped. The mic was probably 6" to 12" away with a pop shield in front of it. I did notice that I needed to back off of it at times because even with the gain set really low it had the tendency to pick up some really gravelly undertones in my voice. About how far do most people record from their mic with regards to vocals? I'm a choral singer, so typically we don't use microphones, just our natural resonance and the environment for reverb.
 

na4a4a

Outwardly Opinionated and Harshly Critical
Supporter
Defender of Defoko
I edited my previous post but feel free to ignore that.

You could probably try standing a little further back from the mic than you already are. Like the pop filter should be at least 5 inches from the mic and you should be about 1.5ft from the mic (not exact but a good place to start, you can use a ruler or even your hands/fists to roughly get the distances).
Any closer than...say...8 inches and you risk proximity effect. Depending on the mic and voice that can vary from a pleasant low end boost to a generally really nasty tone.
 

bearhack

Disciple of Aoede
Supporter
Defender of Defoko
Thread starter
Yeah, I wondered if the proximity effect was going to be too much especially with my tendency to lean on my natural resonance when I sing. Thanks again Jeremy
 
Similar threads

Similar threads