Subject:Getting a good level and crisp audio onto CD

Posted by: kfenner

Date:2/11/2003 4:57:58 AM

I use SF6 to work on Brass Band recordings I take from Live concerts. My normal operation is to isolate the music from audience noise front and back by selection and deletion, adding 2 seconds of silence front, 1 second to rear for each music track. Then I normalize to -16db RMS for music, then run Clip restoral to -6db from Noise Reduction2, then enhance/smooth at a setting of +3 and finally add a little reverb (selection is Bright Hall/Concert Hall). I then resample down to 44.1khz from 48khz (audio is taken from DV video as I record the concerts) with anti-alias filter applied. Then I resmooth at a setting of +3 before creating a CD from the various tracks of the concert.
The issue is, the band members state the final CD needs the volume turning up more on their HiFi's than pre-recorded CD's and pre-recorded CD's sound just that little crisper.
Is there something I could try to overcome these two minor criticisms (these CD's are sold to the band to help fund the band - they are not sold to the public (yet!)).
Thanks
Ken

Subject:RE: Getting a good level and crisp audio onto CD

Reply by: captn_spalding

Date:2/11/2003 9:22:02 AM

What kind and type of microphones are you using? How many mics are you using? Where are they being placed in relationship to the musicians? What kind of mixer are you using? You might get better results by getting the best recording you can up front and do less processing on the back end. Your clients are probably used to listening to highly compressed audio from pop CD's. You may have to decide if you want a recording that is more faithful to the original concert or a more comercial sound that most people have grown used to and consider the norm.

..spalding

Subject:RE: Getting a good level and crisp audio onto CD

Reply by: kfenner

Date:2/11/2003 10:05:32 AM

I'm using Sennheiser mics and a Behringer mixing desk. The mics are not close to the band as they are condenser types (not dynamic) and the band is LOUD when they play. I ensure I get no clipping on the recording of the performance and the recordings are only stereo, no 5.1 surround stuff. (Can you put Dolby Stereo AC3 files onto a CD using Acid4 and the 5.1 plug-in? Would this sound better?)
I have not used any compression in SF6 (nor expander options) as I thought keeping it as close to the original was best - but adding the reverb does make the performance stand out a bit more. The book SF Power points out the use of the smoother/enhancer at a level of 3 after resampling, so that's why I do that.
I use the normalize function as sometimes the levels can be a bit down on the recording as I try to keep sufficient headroom to avoid clipping and distortion, thinking that upping later is easier than trying to lose digital clipping from the beginning. I use the clipping restoral to make sure I have no clips in loud bits.
Any further ideas - I'm open to suggestions to try as I'm new to this audio bit - I usually just work on the video stuff - the CD's are a bonus!
Thanks
Ken

Subject:RE: Getting a good level and crisp audio onto CD

Reply by: Chienworks

Date:2/11/2003 11:46:41 AM

I think the smoothing you're using may be excessive. A single pass at +1 is usually all i care for. +3 twice would probably kill a lot of the high end crispness. I also wonder why use clipped peak restoration. Are you getting clipped peaks from the original source? If not, then you shouldn't have any need for this. For volume, you mentioned normalization, but at first you normalize to -16dB. You also mentioned another normalization step but didn't specify what level you normalize to. If you look at a ripped track from a typical prerecorded CD, the peak levels are often -1dB or higher. You might want to normalize to a higher level.

Overall, it sounds like you're doing a LOT of processing, and some of the processing is being done twice. Each time you process the signal you can introduce errors and distortion, and you lose some of the original definition of the sound. Try doing less processing. I think the first normalization, clipped peak restoration, and smoothing are probably all unnecessary. Even the second smoothing step is probably excessive. Book recommendations are often good, but it's always better to use your ears when deciding what things help.

Subject:RE: Getting a good level and crisp audio onto CD

Reply by: RickZ

Date:2/11/2003 12:10:59 PM

I agree that less processing would probably be better.

FWIW, I record choral groups and use the following procedure with SF6 and Vegas4:
1. Edit out or fade applause to a level below the music on each file of the concert.
2. Normalize each file to -1dB on peaks.
3. If there is a lot of room noise, use 1 pass of NR.
4. Open all files for a given concert in Vegas on one track, in time line.
5. Use Track Compressor, pick setting that sounds good.
6. Render from Vegas to 44.1 as one huge .wav file for entire CD.
7. Open this file in CDA5, drop track markers. Burn CD's.

I've found that the Track Compressor is the key to getting a volume level that people like. Otherwise listening in car, you're constantly tweaking the volume knob.

Hope this helps . .
Regards,
Rick Z

Subject:RE: Getting a good level and crisp audio onto CD

Reply by: captn_spalding

Date:2/11/2003 1:53:17 PM

I agree with most of what Chienworks and RickZ have to say.
I tend to start with no processing (other than trimming/fading unwanted applause), burn a CD and listen to it on different systems. Then I decide what is needed and burn another CD and do it again.
I record mostly orchestras, small group ensembles and individual instrumentalists. With that in mind I offer these comments on the RickZ list:

"2. Normalize each file to -1dB on peaks" I tend to view the concert or recital as a unit and think of the dynamic contrast for the entire unit. So when I do any normalization, it is for the entire recording rather than individual pieces or movemnts.

"5. Use Track Compressor, pick setting that sounds good" The operative phrase here is 'that sounds good'I have a great dislike for compression and almost never use it. On the other hand I tend not to listen to classical music in the car just as RickZ says "constantly tweaking the volume knob". I'd rather have the original dymaic balance when I'm at home listening on a good sysem.

"7. Open this file in CDA5. drop track markers.." I work in SF and use markers, turn them into regions, then open in CDA5 having the regions turn ito events and tracks.

In summary, do ONLY the processing your ear tells you is necessary

By the way on the subject of reverb, check out accoustic mirror and in particular the meyerson hall impulse file. It is a marvel! I had to record a Mozart Requiem in an abysmal auditorium last spring. I close mic'd the orchestra, chorus and solosits and ran it thru meyerson hall. The result was very good.

...spalding

Subject:RE: Getting a good level and crisp audio onto CD

Reply by: Geoff_Wood

Date:2/11/2003 9:08:09 PM

The 'clip restoral' bit sounds a bit strange to me. Why not avoid clipping in the first place ? If CDs are to quiet for taste (my prefenece for such music would be pure peak normalisation) try a higher rms setting: -12 seems a common recommendation.

Also regarding all this 'smoothing' etc, better to get it right before recording, with mic selection and placement. Adding 'Bright Hall' reverb then more smoothing also sounds a little odd. How about choosing a reverb that doesn't require that extra process?

Resampling, even if set to 'best' and anti-alias, may degrqade the sound. My preference is to record at 44K1 in the first place, saving time and quality.

geoff

Subject:RE: Getting a good level and crisp audio onto CD

Reply by: musicvid10

Date:2/12/2003 1:09:01 AM

Right. If you're doing CD's, downsampling from 48 to 44 does no good and maybe a little harm, unless you want the archived files at 48.

I won't comment on the smoothing and Bright Hall mirror, they're just not my thing. I also don't do RMS normalizing. I use a combination of moderate compression and manual surgery on the outliers, final step is peak normalize to -1db.

BTW, the unofficial industry standard is 97% peak, about -.27db, but I again stay a bit conservative.

Subject:RE: Getting a good level and crisp audio onto CD

Reply by: kfenner

Date:2/12/2003 3:25:24 AM

Guys,

Many thanks for your hints. I'll take them on board and try implementing some or all of them in the next cut of a CD from existing concert tracks. The band have asked me to put together a CD of their best tracks for possible sale to the public (now I've to get involved in licencing as well!), so I'll be interested to try out these suggestions and get their feedback.
I'll let you know how I get on.

regards
Ken

Subject:RE: Getting a good level and crisp audio onto CD

Reply by: ELK

Date:2/14/2003 4:45:00 PM

I do precisely this same type of recording, although I am recording a brass quintet and a brass choir (four trumpets, 4 horns, four trombones, tuba). The brass choir does not get as loud as a brass band, but there are many of the same challenges.

First, good mics and recording equipment are critical as in all recordings. (I use Shure KSM44 large diaphragm condenser mics). Good mic placement is also critical.

The trick seems to be to find the balance between close miking (sense of immediacy, great dynamics, but can sound overly bright and balance between sections and within sections are difficult) and miking at a distance (good balance and blend, warmer sound, but can be muddy, unexpressive and lacks power).

Brass musicians are used to power, control and big contrasts. The effect the musicians experience is hard to capture. In a fairly reverberant space, I place the microphones approximately twenty feet in front of the musicians, eight feet high in an X-Y configuration, 135 degrees between the axis of the mics, cardioid mode. I record direct to an Alesis Masterlink at 96/24 or 88.2/24.

I believe normalization and level settings should be done on the basis of the entire concert, not track by track. I usually normalize to -16dB RMS compressing the peaks only if absolutely needed. You could go to a higher RMS level (such as -10dB) if you really need a sense of overall volume, but you will kill the sound of brass which needs uncompressed attacks to retain energy.

I do no other processing other than to lower the overall level by 1dB to allow the necessary headroom to add dither when burning the CD at 44.1/16.

I suspect that all of the smoothing, adding of reverb, etc that you are doing is sapping the energy of the performance and reducing the harmonic complexity of brass.

Good luck!