Forum

Please consider registering
guest

Log In

Lost password?
Advanced Search

— Forum Scope —




— Match —





— Forum Options —





Minimum search word length is 4 characters - maximum search word length is 84 characters

The forums are currently locked and only available for read only access
Processing the voice
Jun 15 2004
11:00
andromeda
Member
Members
Forum Posts: 188
Member Since:
Feb 27 2004
Offline

I know some of you on here have various bits of studio equipment, and some will have had experience of working with real singers. I was just wondering about processing the voice before recording. Most of the units you can buy are designed to process a mic level real voice. Processing Lola is different because a) she is not real b) the signal is line level from the computer. Has anyone experimented with the best way to do this? I know it’s possible to process the raw wave file output too, and iron out the bad bits. I’m thinking more about comparing with the way you would work with a live singer. I have mostly worked with instrumental music in the past and have tended to treat the Vocaloid output as just another instrument. Any thoughts……..
Chris

Jun 15 2004
11:39
drum
Member
Members
Forum Posts: 8
Member Since:
Jun 9 2004
Offline

Not sure how relevant this is to Vocaloid, but there are a couple of tricks that I use when recording real vocalists. Actually, having thought of it I may give it a try when I get home tonight.

This is all done on a Yamaha AW16G 16 track DAW, but I would imagine it’s also possible on a PC/Mac/Whatever.

1) record the vocal track

2) copy this track to two other tracks offsetting +12 and -15 ms respectively.

3) pan the copid tracks to 3 and 9 o’clock

4) pitch shift the copies 3 and 5 % respectively

5) add some reverb

6) bounce them, together with the original track, with the original panned slightly offcentre, to a stereo pair.

Bob’s your aunties live-in lover.

Jun 17 2004
18:24
moog
Member
Members
Forum Posts: 4
Member Since:
Jun 17 2004
Offline

Vocaloid is just another instrument. As you say it’s not a real vocalist. None of the problems of recording a real vocalist are present. Therefore the vocal processors you are talking about are not relevant. Their main function is to solve problems created by recording process.

1. Compression – Not really required with Vocaloid since the dynamic range can be controlled from Vocaloid itself and the possible dynamic range is by definition the same as the recording medium.

2. De-essing – Can’t imagine Vocaloid needs to be de-essed.

3. EQ – Use DAW plugin/Mixer instead – post recording.

4. Noise gate/Expander – Not required with Vocaloid as it doesn’t have background noise caused by the tape or mic.

With real vocals it’s not a good idea to process the voice too much before recording as once the effect is burned into the recording you can’t change it.

Most of the vocal processing can and in my opinion should be done post recording during the mixdown anyway. This includes noise gating, eq, compression and de-essing.

The only processing that should be done pre-recording really is compression, and then just enough to squeeze the signal into the dynamic range of the recording medium. This isn’t an issue with Vocaloid though.

Once recorded here’s some common effects to use on vocals. Of course you can use almost any plug in; just play around and see what happens.

EQ – Just a little to sweeten it up or go extreme from effects. Cut the treble and the bass, leave just the midrange for a common singing-down-a-telephone effect.

Reverb – essential to make the voice sit in the mix. Lock the decay time in with the tempo and have a small predelay (~65ms but lock with tempo) to avoid the reverb muddying the vocal.

Delay (Synched to tempo) – not too loud, not too much feedback – try quater notes or half notes. Can be great at almost imperceptable levels.

Chorus – increases the richness of the vocal

Double tracking (as described drum above) – Makes it sound wide and like there’s more than one singer.

Overdrive/distortion/valve amp – can warm up the vocal with a little or use lots for extreme effects.

Paul

Jun 17 2004
21:41
administrator
UK
Admin
Forum Posts: 268
Member Since:
Feb 4 2004
Offline

I have a few thoughts…

The original recordings of the singers that vocaloid is modelled from, were completely dry with no ambience, processing or compression whatsoever….

So my view is that compression usually is required on vocaloid. I have often found that if I dont compress it, it kind of drowns in the music…. words dissappear in the mix…. some consonants become inaudible. (obviously this depends on the context)

I have often set up a compressor with a slowish attack time. This lets through the starts of syllables and consonants much better. This can be tricky to set up sometimes, so what I often do is use quite extreme settings for ratio and threshold while I tweak the attack time so just the beginning of syllables and consonants like s, t and d are getting through. Then I return the ratio and threshold to a more “normal” setting.

Another thing I find is that sometimes vocaloid sounds too thick lower down (ooh err!) and kinda muddles the sound, with different instruments fighting for the same area of the mix. I sometimes use a multiband compressor aswell just to balance up the frequencies in the voice to the rest of the mix. I suppose you could use an EQ to similar effect. but with a multiband compressor you can control the dynamic ranges of specific frequencies as well as the volume which can be very useful.

A reduction and compression of frequencies between 100Hz and 1K centred at around 250hz can often enable the voice to sit on top of the mix rather than wading through it, which can be tiring to listen too. Obviously this depends on the song, reduction of just a couple of dB is enough.

Also the frequency range 1K-4K is important as it is where most of the clarity of diction in the human voice happens, so sometimes this range will need adjusting.

Sometimes its good to use 2 reverb units. One for ambience and one for actual reverb…. I find you can get better control over those 2 elements that way.

I am sick of typing now

Jun 18 2004
00:58
gray
Member
Members
Forum Posts: 304
Member Since:
Feb 27 2004
Offline

Well if you use more than 1 finger ro type ir wouldn’t take so long, but the typos would increase. Percent typos = # fingers squared X 10
I’m a mathematical genius ya know.

Jun 21 2004
07:23
andromeda
Member
Members
Forum Posts: 188
Member Since:
Feb 27 2004
Offline

Thanks guys for all the replies. You are right (Joe) about Lola being a bit heavy in her lower region!!!! I’ve just been buying some more outboard equipment to beef up my system and the advice will help me to get the best out of it and to work out the best way of connecting and routing the signals. One question, Joe. What microphone(s) did you use to record Lola’s voice?
Chris

Jun 22 2004
04:07
mrdc
Member
Members
Forum Posts: 4
Member Since:
Jun 18 2004
Offline

:?: Do any of you know what mic’s were used to record Leons voice? One thing I have play’d with is multiband compression too, but then I add a doubler effect. I am creating in Nuendo 2 and have always used the automation to ride the vocals, I find this works great with Leon too as I don’t really want to spend time in the vocaloid editor doing that beyond setting the velocity,,,so this kinda saves me time. The trick is he is too perfect and its hard to get him out of time a bit here and there to make it more human sounding > but thats a bit silly isn’t it LOL!
Later.
MRDC
:D

Jun 22 2004
07:13
andromeda
Member
Members
Forum Posts: 188
Member Since:
Feb 27 2004
Offline

I agree. My techique to achieve some realism is to a) detune the voice and b) move the start of the notes away from their grid positions in the Vocaloid editor.
a) use the pitch bend parameter on individual notes so that they sound ever so slightly flat or sharp. This is over and above any pitch bend used as part of the melody.
b) The best way to get the rhythm sounding right is to play in the melody from a keyboard – without quantisation. Create a midi file and import it into the editor. The melody you record will have natural errors in timing.
Chris

Forum Timezone: UTC 0

Most Users Ever Online: 108

Currently Online:
4 Guest(s)

Currently Browsing this Page:
1 Guest(s)

Top Posters:

gray: 304

robotarchie: 223

andromeda: 188

quetzalcoatl: 173

Giuseppe: 167

Luka Mitutoyo: 147

Member Stats:

Guest Posters: 1

Members: 7988

Moderators: 0

Admins: 1

Forum Stats:

Groups: 5

Forums: 27

Topics: 764

Posts: 3555

Newest Members: kprem, yang7764, lisajim, gameace, Sunnydoot, Anildoot

Administrators: administrator: 268