Recording levels in Sound Forge etc at DVinfo.net

Jim Andrada · July 14th, 2008, 03:48 AM

I've been recording dialogue with an SD302 mixer into an SD702 recorder.

All the levels look good on the 302 and the 702, and the sound in the headphones is fine both when recording and on playbeack in the 702

When I bring the file into Sound Forge, the levels look really really low and I have to crank up the gain on the speakers to hear anything. If I increase the gain by 20dB in Sound Forge, the peaks seem to be right around -6dB and it sounds OK through the speakers.

Any idea what's going on and why the levels seem so low in SF???

David Beisner · July 14th, 2008, 09:30 AM

I've found I often see that problem when I'm recording, especially going to a digital recorder. It's really hard to maximize the input level on a digital recorder without also peaking out, and as a result we tend to unconsciously set the levels a little lower than we should and compromise by turning up the output of our headphones. I'd recommend putting a compressor/limiter in your chain if you're able. That way you can get the full dynamic range possible without clipping and distorting your audio. Digital distortion due to clipping is impossible to remove in post, so better to be too quiet than not loud enough.

If you're having trouble amplifying your audio that was recorded too low without getting hiss or other white noise, try amplifying using an EQ rather than an amplification plugin. That will allow you to specify the frequency ranges that get amplified, reducing the amount of amplication that happens in the problem frequencies up high and down low.

Steve House · July 14th, 2008, 10:03 AM

Sounds like improper gain staging to me. Do not use the phones to set levels, use the meters, that's why they're there. Send OVU tone from the 302 to the recorder and set the recorders inputs so tone reads -20 dBFS. Engage the 302's limiters. As you record adjust the 302's input faders so the audio peaks around 0VU or a little higher on the 302. That should give you adequate levels with about 18dB of headroom. If the source doesn't have wide excursions in level you could "tone up" at -18dBFS or even -12dBFS to record at a higher average level.

Jim Andrada · July 14th, 2008, 06:58 PM

Well, after playing with it some more I think I have an idea what's wrong. Mixer/recorder were set as suggested, but I think I ws a bit"chicken" to set levels high enough. I mainly record groups like brass bands where you need to leave "ample" headroom. With a straight moderate volume normal speaking voice there jud are't such high amplitude excursions to worry about. -12dB worked just fine.

Ty Ford · July 14th, 2008, 07:14 PM

Yes, bad audio frequently tastes a lot like chicken. :)

Regards,

Ty Ford

Jim Andrada · July 14th, 2008, 07:37 PM

Or maybe like chicken @@@@ (And I don't mean chicken tenders!) :<)

Actually, the dialogue sounded quite good after bring boosted - better than I deserved!

Oh well - live and learn!!!

Ty Ford · July 14th, 2008, 08:06 PM

Jim,

When I reviewed the 744T (big brother to the 702T), I was asked by Sound Devices to purposely record a track with peaks at -30 and normalize to hear how good the converters were. Yes, there was some noise, but the track was usable. That's scary good.

Regards,

Ty Ford

Jim Andrada · July 14th, 2008, 09:58 PM

I was quite surprised how well the 702 did in that regard - I think my peaks were around -20 and the result was better than just usable, it was rather nice. I think this says a lot about the value of low self noise in the whole recording chain.

Regardless of which, I'm paying a lot more attention to getting it right in the future. Lesson learned!

July 14th, 2008, 03:48 AM	#1
Jim Andrada Inner Circle Join Date: Feb 2007 Location: Tucson AZ Posts: 2,211	Recording levels in Sound Forge etc I've been recording dialogue with an SD302 mixer into an SD702 recorder. All the levels look good on the 302 and the 702, and the sound in the headphones is fine both when recording and on playbeack in the 702 When I bring the file into Sound Forge, the levels look really really low and I have to crank up the gain on the speakers to hear anything. If I increase the gain by 20dB in Sound Forge, the peaks seem to be right around -6dB and it sounds OK through the speakers. Any idea what's going on and why the levels seem so low in SF???

July 14th, 2008, 09:30 AM	#2
David Beisner Major Player Join Date: Jun 2008 Location: Dayton, TN (USA) Posts: 219	I've found I often see that problem when I'm recording, especially going to a digital recorder. It's really hard to maximize the input level on a digital recorder without also peaking out, and as a result we tend to unconsciously set the levels a little lower than we should and compromise by turning up the output of our headphones. I'd recommend putting a compressor/limiter in your chain if you're able. That way you can get the full dynamic range possible without clipping and distorting your audio. Digital distortion due to clipping is impossible to remove in post, so better to be too quiet than not loud enough. If you're having trouble amplifying your audio that was recorded too low without getting hiss or other white noise, try amplifying using an EQ rather than an amplification plugin. That will allow you to specify the frequency ranges that get amplified, reducing the amount of amplication that happens in the problem frequencies up high and down low. __________________ David Beisner Media Specialist, Bryan College, Dayton, TN -- www.bryan.edu

July 14th, 2008, 10:03 AM	#3
Steve House Inner Circle Join Date: Mar 2005 Location: Hamilton, Ontario, Canada Posts: 5,742	Sounds like improper gain staging to me. Do not use the phones to set levels, use the meters, that's why they're there. Send OVU tone from the 302 to the recorder and set the recorders inputs so tone reads -20 dBFS. Engage the 302's limiters. As you record adjust the 302's input faders so the audio peaks around 0VU or a little higher on the 302. That should give you adequate levels with about 18dB of headroom. If the source doesn't have wide excursions in level you could "tone up" at -18dBFS or even -12dBFS to record at a higher average level. __________________ Good news, Cousins! This week's chocolate ration is 15 grams!

July 14th, 2008, 07:14 PM	#5
Ty Ford Inner Circle Join Date: Nov 2004 Location: Baltimore, MD USA Posts: 2,337	Yes, bad audio frequently tastes a lot like chicken. :) Regards, Ty Ford __________________ Ty Ford's Audio Bootcamp Field Guide Ty Ford's Audio Video Blog

July 14th, 2008, 08:06 PM	#7
Ty Ford Inner Circle Join Date: Nov 2004 Location: Baltimore, MD USA Posts: 2,337	Jim, When I reviewed the 744T (big brother to the 702T), I was asked by Sound Devices to purposely record a track with peaks at -30 and normalize to hear how good the converters were. Yes, there was some noise, but the track was usable. That's scary good. Regards, Ty Ford __________________ Ty Ford's Audio Bootcamp Field Guide Ty Ford's Audio Video Blog

July 14th, 2008, 06:58 PM	#4
Jim Andrada Inner Circle Join Date: Feb 2007 Location: Tucson AZ Posts: 2,211	Well, after playing with it some more I think I have an idea what's wrong. Mixer/recorder were set as suggested, but I think I ws a bit"chicken" to set levels high enough. I mainly record groups like brass bands where you need to leave "ample" headroom. With a straight moderate volume normal speaking voice there jud are't such high amplitude excursions to worry about. -12dB worked just fine.

July 14th, 2008, 07:37 PM	#6
Jim Andrada Inner Circle Join Date: Feb 2007 Location: Tucson AZ Posts: 2,211	Or maybe like chicken @@@@ (And I don't mean chicken tenders!) :<) Actually, the dialogue sounded quite good after bring boosted - better than I deserved! Oh well - live and learn!!!

July 14th, 2008, 09:58 PM	#8
Jim Andrada Inner Circle Join Date: Feb 2007 Location: Tucson AZ Posts: 2,211	I was quite surprised how well the 702 did in that regard - I think my peaks were around -20 and the result was better than just usable, it was rather nice. I think this says a lot about the value of low self noise in the whole recording chain. Regardless of which, I'm paying a lot more attention to getting it right in the future. Lesson learned!