Recomended Audio Level (in db) for DVD recording

[BJ_M]

see my response in your other thread

[hiptune]

does this whole issue have anything to do with the post I just made today in this forum? The topic is called:

? for audio gurus ! ! (recording guitar)

absolutely not a thing!

[SaSi]

No, this thread is about a guy making a simple question, getting a couple of responses, being happy with them, and then other people jumping in exchanging gibberish for passtime...

[Gildas]

I capture a lot of video from my DirecTV TiVo using my iLo DVD recorder. I rip the DVD to my hard drive and then do some editing (mostly to remove commercials).

I use Goldwave to edit audio files and I use the Normalize function a lot. I always wondered what the reference was for the dB values used in Goldwave. For information, when I open the audio part of a captured video, the volume is at a level of about -24dB on average. Using Goldwave I normalize it to about -16dB and the sound is quite loud when I play the DVD on my TV, which is what I wanted.

All this to say that -16dB is a good value for me and gives me a sound louder than regular TV broadcasts.

I'm glad I read this thread because I learned a lot of things related to different dB units of measurements.

[Cornucopia]

The goal of good audio recording is to get the best fidelity that you can. That usually means "Maximum Dynamic Range", while remaining as CLEAN as possible (no distortions).

In Analog recording, where the machines had improving Signal-to-Noise ratio (up to a point of diminishing returns) and very slowly increasing Distortion up to a point where it would start to rising dramatically (Saturation, ref. to tape), you kinda knew what to do. These points were close together. As long as the great majority of the recording hovered around a point at or below the S/N point ot diminishing returns, with the peaks hovering at or below the saturation point, that was as good a signal as you could get. Now, there are various types of audio material with varying dynamic ranges, specifically the range between the RMS average level, and the highest peak level. For recording purposes, it's best to set the RMS at the 0VU recording level (either referenced to +4 or -10 depending on your system), and the peaks below saturation--assuming you have enough headroom (distance between 0VU and Saturation point)for that. If you don't have enough headroom, you lower the recording level. Basically, you're working with two points of reference

Now, how does that translate into DIGITAL recording?
0dBfs = 0dBfs, no matter what system you're using. Above will clip, below is clean. S/N (or more accurately Signal-to-Error) is smoothly linear--closer to 0db is better--but don't clip! Unfortunately, many systems choose different criteria for telling you WHEN you've clipped. Some will count 1 0dBfs sample as a clip, some 2samples-in-a-row, some 3, some much more. What they're doing is guessing what you might be using as a source signal. If you had a square wave recorded cleanly at 0dBfs, to the system it could look like EVERY sample (except the zero-crossing ones) were clipped. So they make a guess, giving you a little room for error.
Then, you also have to adjust for what type of signal dynamic range you're giving it, because some transients may sound fine even when clipped 20-in-a-row, while most might be glaringly problematic.
The point here is to keep your OVERALL signal as close to 0dBfs as possible, while not screwing with your dynamic range too much. That means, if you have a signal with an AVERAGE RMS level of -16dBfs, with peaks that go up 15dB above that, you are running the overall peak level at -1dBfs--very good level. But if your signal had peaks that went 24dB above average, you'd have clipped off 8dB worth of signal--very bad! So you can either apply compression/limiting to the signal, to get those peaks back down below clipping (BEFORE RECORDING), or you can just lower the overal level by 8dB. That's why there are various "standard" recording levels. If you work with TV News/Speech/LowDynamicRangeMusic(Pop/Rock/Rap), you'd set your levels at -10/12/14/15dBfs. If you do LiveRecording/Jazz, you might set your levels to -18/20dBfs. If classical/movie/symphonic/avantegarde (where you never know how loud something will peak) or if you do recording of natural sound effects, you may want -22/24dBfs. Those are good, professional starting points.

Note that I didn't say anything yet about NORMALIZING. Normalizing isn't something you do during recording, it is a BAND-AID for when you don't get the level right in the 1st place. It is also not always the best solution. Whenever you raise the level via normalizing, your are also normalizing the Noise and Quantization Error (heard as distortion). You gain only maximum output level, but don't gain ANY dynamic range. You could do the same by raising the volume knob up--normalizing just automates the process and permanently saves to file. Normalizing also ADDS distortion, unless it's super-high quality and includes dither.
If you're recording the right level, you'll hardly ever need to normalize.
If you've recorded too low, you could always raise levels via a compressor/limiter/dynamicrange plugin (one that includes dither) and this would probably make your program less "gritty" than normalizing would. If you've recorded too low, you should do a simple, dithered gain change (loss).
How close do you need to be? Remember that every bit (of the 16/20/24 bits) corresponds to 6dB of S/N. If you are peaks are within less than 6dB, then you may not want to do anything (that's less than changing 1 bit's worth). Also remember that great high quality analog recordings had ~70-80dB of dynamic range, so if you started out using 16bit and recorded your PEAKS at -18dBfs, you still are at 78dB of dynamic range (it's even better at 20/24bit). So you have more room to fudge--as long as you have good A/D converters, and apply gain/dither correctly and appropriately.

My $.02,

Scott

[sdsumike619]

that was more like your $10.00
But thanks for it =)

[hiptune]

I just listen to several takes, and make choice which has nicer sound, loud without distortion.

I would never normalized a digital sound track ever. And I have no need to try to make my dvds louder than my TV or other video content.

Get the sound captured with good levels, and leave it alone.

[Gildas]

You make a good point, hiptune, when you say to capture with good levels but the problen is that some channels have a lower volume than others so it is not my choice. I can't set the level of the recordings between my TiVo and my DVD recorder to correct for the difference between channels so I have to use the software tools that I have.

It works for me and I am happy with the results.

To each his own.

[hiptune]

I am not sure what you are doing, and if you are preserving quality when you rework the video and audio in the DVD recorder.

At any rate, this stuff is so much work and effort to get right, that I can't stand any degraded picture or sound. And I would find out a solution that gets the volume level up to where you want it before you capture. I am only capturing things that you can't buy and likely never will.

I am sure there is a way to boost audio, and it may involve capturing a different way. I am also of the thought that is if the sound was loud enough for the tv broadcast to begin with, how bad can it be playing off a dvd later?

But if it really is to each his own, then there are no good reasons to post here. Just go on and do your own thing. The rest of us that like to learn what others are doing, and believe there are other ways, can continue to seek others advice.

To you, and your own.

[carroll_ray]

On a somewhat related note, I'm currently trying to capture some VHS tapes (mainly tapes of family reunions, things like that) and burn them to DVD using my ATI All-In-Wonder Radeon 9200 (I believe that's the model number). In any event, I'm having trouble with the audio portion of the capture. While doing the capture, the audio sounds just fine. However, during playback, it's clipped/overdriven. The audio is coming through the ATI-supplied "breakout box" to the line in connection on my onboard sound card (Asus MB with an NVidia sound chip set).

Seems that no matter what I set the volume on the record side of the sound "card" to, the audio portion of the capture is overdriven/clipped. Can someone point me in the right direction to correct this problem?

I may end up having to spring for a standalone DVD recorder, and just do a straight dub from VHS to the DVD recorder. I was hoping, however, to be able to do it with what I've already got.