|
VO-BB - 19 YEARS OLD! Established November 10, 2004
|
View previous topic :: View next topic |
Author |
Message |
soupy Guest
|
Posted: Sun Apr 24, 2005 6:54 pm Post subject: Compression & Normalizing - Telephony Voice overs. |
|
|
Hi,
When recording for telephone on hold / IVR projects I normally just:
(Using Sound Forge).
1. Record in at comfortable input gain by mike with phantom power. Have a neat firewire recording interface.
2. Then at editting / mastering stage the main process I apply is 2:1 compression and bring up the gain in that process too if need be. If I am mixing with another voice then gain would be matched up.
My question is should I be doing another process in addition or instead of 2:1 compression in the final mix?
i.e wavehammer, normalizing.
I have found 2:1 is most suitable for typical situation for on hold projects.
Some advice would be appreciated. |
|
Back to top |
|
|
TomRVO Guest
|
Posted: Mon Apr 25, 2005 6:08 pm Post subject: don't think you need compression |
|
|
hi, soupy,
my experience is that telephone lines are heavily compressed at it is. i suspect that all you'd need would be to normalize it.
tom |
|
Back to top |
|
|
Frank F Fat, Old, and Sassy
Joined: 10 Nov 2004 Posts: 4421 Location: Park City, Utah
|
Posted: Mon Apr 25, 2005 7:55 pm Post subject: |
|
|
Normalizing is a great step in the right direction. There are a couple of variables when it comes to telephone lines which may interest you.
Is this "dry voice" or "mixed with music"? If dry voice - a slight touch of compression may be desireable - a very slight amount though goes a long, long, long way on telephone lines. IF mixed, take your music down to being - just audible - raise that 10% and you have a perfect mix for the telphone. Any louder than that and the music will drown out the spoken word. When not speaking - you may raise the music another 10 to 20% - but not much more.
Second: the "female" voice is very different from the male voice when it comes to phone lines and compression - it's a touchy-feely thing. The female voice will sound much louder on a phone line recording than a males when recorded at the same levels - so - if you have a female voice take her sound level down approximately 30%/-3db below yours and you will sound like you were speaking at equal levels.
Additionally: as you are using Sound Forge, Wave Hammer may be used, just make sure you normalize BEFORE and AFTER to approximately 85% as your peak level. In "Hammer" use Soft Compression and set for a very minimal compression settings to get the best sound. (Threshold: -8.0db, Ratio, 2.5:1- or lower, 0.0 db Output gain, and then normailze to 85%).
Good luck,
Frank F |
|
Back to top |
|
|
soupy Guest
|
Posted: Mon Apr 25, 2005 10:40 pm Post subject: Compression & Normalizing - Telephony Voice overs |
|
|
This is interesting Frank.
I am referring to the context of mixing voice with music.
I never add anything further to the music at all. That why processing is applied to voice and not to final mix with music. Otherwise music gets a lot more brighter and punchier than it needs to be.
Processing is only added to dry voice prior to mixing with music.
It is interesting to note the Female voice at same gain will sound brighter than the mail voice at same gain. Its not something I have been critical with measuring. Normally I just measure the output gain onto an anlog vu meter and line up both voices that way.
I sometimes compare the wav shapping / sizing in sound forge for further reference.
Music needs to be about 30% less than the voice gain otherwise it gets too loud.
In summary should I focus your smooth compression wave hammer process instead of the 2:1 compression on its own?
Also what is the advantage of normalizing after adding compression IF the gain is already at the appropriate level as a result of compression to the desired gain as well?
sorry if I am sound a little long winded. |
|
Back to top |
|
|
Frank F Fat, Old, and Sassy
Joined: 10 Nov 2004 Posts: 4421 Location: Park City, Utah
|
Posted: Tue Apr 26, 2005 1:24 am Post subject: |
|
|
No problem.
To quote you: "Music needs to be about 30% less than the voice gain otherwise it gets too loud" This is essentially correct, due to the codec properties of a telephone transmission (highly compressed) certain frequencies will become, as you say - "brighter". Sustained notes, as in music, are essentially processed differently.
"...should I focus your smooth compression wave hammer process instead of the 2:1 compression on its own?" The answer is yes. The "Smooth Compression" will give a cleaner overall compression/limiter value for certain frequencies - especially valuable with the limited frequencies available on a telephone codec. Note: VOIP telephones have different codecs available which if the originating telephone is on VOIP - it will change how the reciever's decoder will "hear" the sound.
"...what is the advantage of normalizing after adding compression IF the gain is already at the appropriate level as a result of compression to the desired gain ..." In normal circumstances you would record your voice at a level less than 75%, then do your editing and mix, then normalize, then "Hammer" it, then normalize again to keep the levels to approximately 85%. Using the settings as I described previously, the compression of Wave Hammer would set a level of about 100% to 107%. Remember, a phone has a limited frequency range and bandwidth to work with, so, to give a bit more "head room", and add a little more dynamics in the processed sound of a telephone; "normalize" to a level of approximately 85% and you will hear the results.
Leaving your music or sound effects UN-compresssed prior to the final "mix", is a great idea for IVR. Just remember, when you "normalize", you bring up the levels of the music/SFX as well. Make your mixed music/SFX so they are just above - "almost audible", and you find in the end - it just sounds better.
I am trying not to be too technical here, so if I have skipped the nitty-gritty and glossed over this - that is why.
Ours is not to understand, ours is to just "do" - and make it sound pretty.
Frank F |
|
Back to top |
|
|
soupy Guest
|
Posted: Tue Apr 26, 2005 3:03 am Post subject: getting somewhere. |
|
|
Ok I think I am getting someware now.
So we should always normalize RMS and NOT PEAK for both the normalize & wavehammer function?
So according to your steps the output gain of each process should result in:
Initial gain approx 75% (raw recording un processed).
1st normalize -> gain rise approx 10% = 85%
2nd wavehammer -> gain fall approx = 75%
3rd normalize --> final gain = 85%
Whats the best way to measure these gain changes accurately to make sure it is not rising or falling too great at each step? by looking at level meters or viewing size changes of the wav?
My initial test shave resulted in the voice being less deep, with more mid range fidelality and I guess clearer which is the idea.
I forge to mention, the processed voice is sequenced in ACID where gains of each music track are adjusted there.
Frank you sound like a real sound engineer from way back. I'll get there one day... |
|
Back to top |
|
|
Frank F Fat, Old, and Sassy
Joined: 10 Nov 2004 Posts: 4421 Location: Park City, Utah
|
Posted: Tue Apr 26, 2005 7:29 am Post subject: |
|
|
Soupy, email me privately for more. I believe I can help you with more that way.
Frank F |
|
Back to top |
|
|
kgenus Seriously Devoted
Joined: 01 Dec 2004 Posts: 889 Location: Greater NYC Area
|
Posted: Tue Apr 26, 2005 8:38 am Post subject: |
|
|
Have you monitored the mix through the phone patch to your monitors? It slows things down a little, but you get a good idea of the process based on the particular phone system.
Kevin _________________ Genus |
|
Back to top |
|
|
|
|
You cannot post new topics in this forum You cannot reply to topics in this forum You cannot edit your posts in this forum You cannot delete your posts in this forum You cannot vote in polls in this forum
|
Powered by phpBB © 2001, 2005 phpBB Group
|