I just bought this Zoom H1 recorder and I wanna ask whats the ideal input audio level. There is a scale from 0db to -40db, where should I keep my levels? On -12db? Thanks for any advice!

I assume your are asking about meter display levels.
Simple: As close to 0dB as possible. Not so simple: Without exceeding 0dB. (aka, clipping)
So generally, a average meter reading hitting around -12dB with peaks 'occasionally' hitting -6dB or so, is usually good. You can be more conservative using 24 bit depth. For instance. 48kHz / 24bit uncompressed PCM files. (.wav) .

Careful, there's "average" and then there's "averaging." An averaging meter (VU meter) doesn't show the speech short-term peaks very well so when it is reading about -20dB the peaks are actually hitting closer to around -12dBFS and will read that level on the peak-reading meter typical of most digital devices. If your Zoom's meter is hovering around -12dBFS to -10dBFS on regular speech you're pretty close to optimum with a good safety margin before you risk clipping.

Yes, I'm talking about meter display levels.

What is dBFS?

I should have used the word "typically" instead of average. Sorry for any Peak vs. Average confusion.
What is dBFS?
> Decibels, Full Scale . Usually noted as dBFS . 0dB being the absolute maximum. And that is what the H1's meter displays.

Yes, I'm talking about meter display levels.

What is dBFS?

Decibels Below Full Scale. Meters for digital devices are calibrated and numbered with the maximum the signal can reach without clipping at the top or full right-hand end of the scale, representing the level where the digitized signal is all ones and so no larger number can be expressed. Signal levels are measured in terms of how far below that level they fall. -12dBFS means you're 12 dB below the clipping level while -20dBFS means you have a weaker signal 20 dB below the clipping point. Because of the nature of meters, the instantaneous signal level can be as much as 3dB higher than what is showing on a meter, simply cause the meter can't respond fast enough to indicate the level of a single cycle in the waveform. That's why you don't want to record with the meter reading hovering all the way up near the top. Plus a normal dialog scene will have characters speaking at different levels from moment to moment - if you've set the level so the meter reads, say, -6dBFS when he's talking in a normal voice and suddenly he yells or claps his hands, that high level could touch the clipping point. -12dBFS is a good compromise between recording as hot as possible in order to obtain a good dynamic range compared to the device's noise floor, with a bit of cushion to absorb sudden loud sounds and a safety for the unseen occasional higher peaks as well.

Good explanation. Thanks.

BTW, should I somehow adjust the levels in postproduction before uploading the video to the internet or should I leave it at -12db?

Very good question!

Broadcast seems to be at -12dB and I tend to leave audio for DVDs at this level as well. (Too high a signal and you can get domestic audio gear crackling and distorting - not worth it for the complaint factor.)

For internet delivery I'm personally peaking it at -6dB which keeps it nice and "strong" for the end user and their computer speakers, but with a bit of headroom for whatever it may be worth. There really isn't a standard when it comes to uploading to YouTube etc. That's just what I do.

Andrew

Good explanation. Thanks.

BTW, should I somehow adjust the levels in postproduction before uploading the video to the internet or should I leave it at -12db?

Levels of the original recording are rarely the levels used in the final production nor is there any reason they need to be. You adjust either up or down as needed for the final mix. Indeed, you might have several mixes of the same show for SD broadcast, digital broadcast, DVD, and internet, all with their own levels. Think of the original recording as being the negative while the final mix is the print.

Yeah, -6dB is good for the web. Compression/limiting may be required.

I recently transcoded a corporate video that was done at -12dB and was to be shown in a corporate meeting. Two people previewed it. Both complained that the levels were too cold. We then previewed it in the meeting room. The system didn't have enough gain for a -12dB average (speech only) signal. I compressed and boosted it to -3dB and everybody was happy.

No amount of explaining about professional levels would have mattered. The bottom line is that -12dB didn't meet expectations for video viewed from a computer.

Interesting.

Yup, and I would add that I really believe in applying the "broadcast" setting on the multi-band compressor (in Audition) to content that is about to go up on to the web.

Andrew