I just finished a 2 hour interview with two speakers, one of whom pretty much faced his mic and gave me a fairly consistent audio track. The other kept turning her head, so her mic volume is all over the place (If I’d had a lavalier I’d have used it).
Listening to the raw audio I see I’ve got a serious problem. Her volume goes from a healthy -2db to almost nothing. Argh. Not enough of her got picked up by anyone else’s mic to use as a fix.
Fortunately, background noise is near zilch. So if I can find the right tools in Audacity, I should be able to equalize her volume without inflicting a monstrous background hiss. I’ve heard the C/L can be tricky so was wondering if anyone could give me a couple of initial tips on how best to take a mono track, one speaker, with crazy volume variations, and flatten it into something more or less consistent.
I’m not a sophisticated Audacity user. I generally do some simple post processing like Noise Reduction and Normalise, but this is right off the map for me I got everyone set up and then took the headphones off so interviewees would feel more natural, and that was the mistake I guess. Ouch.
Hints? I did google around some, but most of the hits I got were about levelling out volume disparities between N tracks before mixdown, not about fixing wild volume variation w/in one single track.
After writing this, I googled some more and finally hit the right search terms, found this: https://forum.audacityteam.org/t/make-quiet-parts-of-audio-louder/36453/1
If this is the Answer then I’m home free. Off to try it…
I got everyone set up and then took the headphones off … that was the mistake
This is all happening in a mixed show? No individual tracks?
[slowly backing out of the room]
There is one possibility if the voice differences aren’t too crazy. Chris’s Compressor.
Chris is a full-on broadcast compressor which will level out different performers in the same show. I change the first setting, Compress Ratio from the default 0.5 to a stiffer 0.77 and it gives me a show nearly identical to the local radio station. No other changes.
It won’t do hours and hours, so you may need to break up a long show into half-hour chunks. The other problem is the show ends. Chris doesn’t like them. Put some “fake” sounds on the ends of the segment you’re processing and then cut it off later.
You may have produced your way out of the Hollywood Desperation Move. Have somebody in front of a good microphone in a quiet room read the same words.
You can do this free-form although it seems like you can’t. Listen to the live performance in your headphones and repeat the words half to a second later. It takes practice, but normal humans can do this.
Mix your brains out. If there are times where you can’t make out the words, you’re really close to Show Stopper.
If this seems like a lot of work, you’re right. Is there a time limit on show delivery? You’re talking about a retirement project. How old are you?
Can you edit around the missing words? This is totally not going to be done by dinnertime.
LevelSpeech2.ny plugin is worth a try.
Her volume goes from a healthy -2db to almost nothing.
“Almost Nothing” is not going to be visible on the timeline. Most of these “convenient” tools need something there.
Thanks all for the very entertaining (and educational) responses. I think my favourite is Koz “slowly backing out of the room” – lol.
Using the link I found, I managed to get Speaker B back up to almost normal levels, but it looks like Trebor’s LevelSpeech2 does a better job so I might use that for the final pass.
Yeah, we can’t use the studio because Covid, so I’m having to do interviews wherever I can out of the rain (in spaces big enough for pandemic-spaced seating) with a Zoom HN4 and long xlr cables. The HN4 only takes 2 real mics and then you’re stuck with the condensers on the top, which imho are far too sensitive and pick up everything for miles around Long story. Anyway, it’s local community radio not pro stuff, so I can get by with sub-CBC standard
Last one I did was interesting, we did it over Zoom and my interviewees (a couple) had their own Zoom H4N … so I had them record their end, I recorded at my end, and the Zoom (ahem, the video conferencing software called Zoom) recording was just a last-ditch backup that I never had to use.
Much obliged to y’all and hope I put a chuckle (or at least a snicker) in your day…
we did it over Zoom and my interviewees (a couple) had their own Zoom H4N … so I had them record their end, I recorded at my end, and the Zoom … recording was just a last-ditch backup that I never had to use.
You’re getting the hang of this.
Did you develop a semi-silly way of generating sound sync? My favorite is the connection manager says: “And one and two and three and (everybody) clap.” Use the clap(s) to align the first words. Nobody ever thinks of it, but you can do that at the end, too. Nobody is guaranteeing that all sound recorders are operating at exactly the same sampling rate. Fix with Audacity > Effect > Change Speed.
If the goal of a Zoom is a “radio” production, everybody wears headphones. All the Zoom Direction Management, Ambience Suppression, and Echo Cancellation sound errors go away.
I produced my technically successful audiobook test with a Zoom H1n on a stack of bathroom rolls and a quiet room. I suspect people think I’m kidding when I tell them to stop struggling with recording on the computer and use an actual sound recorder.
There is a video of Josh using an H4n complete with cables running everywhere, but he did the whole first part of his music career with an H2 (no “n”).
Those puppies are still going for new equipment prices on eBay.
Interesting, so just plugging in the phones turns off all that annoying Zoom “simplex” stuff? good to know.
We have pretty sketchy bandwidth around here (rural, radios up trees) so Zoom video is touch-n-go and even the audio gets janky at times. I would never use it as my primary record of the interview.
Agree, I totally believe in portable digital audio recorders. USB mixer interfaces meh. Hate the Zoom H series UI, but love the devices. I’ve just invested in my own personal rig, a Zoom PodTrak P4 which is closer to my heart (4 REAL mics, and phones for everyone if I read the product lit right). Gonna give back the station’s gear and stick with my own kit which will be better cared for.
Anyway I just tried a few experiments with Trebor’s Level tool and it looks like a winner…
Funniest post story I have actually is the day I was trying to get good audio for a 3 way interview and used the H4N for the interviewees, then recorded my end with a decent lavalier and an iPad running, oh, some audio journo tool, forget its name, it was OK. So I come home with two good audio files, but when I get them into audacity I discover oh boy, subtle clock variation between the two tools! so my 2 hours of audio, by the end, is like 4 minutes out of synch… and there was just enough bleed through of my voice into their mics, and theirs into mine, that it mattered! the echo effect got weirder and weirder over time from phase shift to complete gibberish. Audacity to the rescue! I could stretch selected passages just enough to get them back into synch.
That taught me to at least use the same brand/model of device if I’m gonna have two of 'em going at the same time. Live and (painfully) learn eh!
There is also old free stand-alone software called the levellator
It was resuscitated in 2015 so that it continued to work on Macs …
just plugging in the phones turns off all that annoying Zoom “simplex” stuff?
There’s a trick to it. You wear headphones so your voice sounds good to everybody else. Zoom doesn’t have to subtract their voice from your room on your outbound voice. That’s why the Zoom Manager has to insist on it. Your headphones don’t change much for you. If anything, they’re a little annoying.
One of the two PBS stations does a brief Zoom news roundtable and the best variation I saw was one newsie who was wearing wired earbuds. Perfect functionality and the only way I know what’s going on is the tiny wire down their chest.