I am very much a newbie in the audio recording arena – getting ready to try my hand at audio book recording. Therefore, I have attached a sample .wav file for your evaluation/critique. As instructed from kozco.com’s site (which is awesome by the way), the file is raw. I used the ACX Check utility and it showed as being acceptable in its raw form, but I’m interested to hear your feedback. Following, if you need them, are the specs on my equipment:
Operating System: Windows 7 Pro
Software: Audacity 2.1.2
Microphone: Audio-technica AT-2035 (with pop filter)
Audio Interface: Focusrite 2i2
Cool. Somebody else with naturally bent blue waves. Did you catch that your blue waves look a little like they got too hot and they melted a little; that you have more blue on the bottom than on the top? My voice is almost exactly the opposite.
It did take a while to figure out that wasn’t a broken microphone. Particularly awkward because I do equipment evaluations. Here, let me check this microphone with my broken voice. What could go wrong?
Do I want to know how you got to -67dB background noise? If it was by processing, I can’t detect it, so we’ll assume ACX won’t be able to find it, either. They hate noise reduction. It’s not unusual for somebody to read in a low volume voice with high noise and then try to rescue it with what I call “Disaster Recovery” tools. I know pros who can do that…
You sound like you’re good to go.
If somebody forced me to find something wrong, you do have the slightest P pop on Natural PPower of your imagination. I easily got rid of it in post production with the custom rumble filter, but you should probably try to not make them at all by pushing your microphone just slightly left or right off center or up, so the sensitive part lines up with your eyes or nose rather than your lips. My personal favorite is having the mic line up about with my left cheek. That position is missing the lip air blast, is almost as loud and many times I don’t need the pop filter.
Your mileage may vary, consult your local listings.
Reading promotion and publicity work is kind of cheating. It’s not all that hard to do a television commercial voice. Try reading an audiobook where the goal is to sound like you’re telling somebody a fascinating story over cups of hot tea on a kitchen table. Oh, and do interpretation of all six characters while you’re there.
I wonder why that grit didn’t show up when I listened to it. I’ll listen again when I get back in the house.
the file is raw.
That’s impressive. I can achieve that level of first-pass quality but it’s hard work. That explains why I couldn’t find any processing artifacts. No processing.
I wonder where that aural grit is coming from… This would be a perfect transfer save that.
I sincerely appreciate the feedback you and Trebor gave me. However, I have to admit that some of it was a bit over my head. When I wrote initially that I’m a “newbie” . . . I meant a still very green, freshly hatched newbie! Ha! Ha! Now that you’ve listened to my sample file, I have a couple of questions. I know where the louder “P’s” are likely coming from. I had to get really close to my mic/pop filter in order to get rid of the RMS values not being acceptable in the raw file.
I have my mic recording level set to the highest (1.0) number, and played with gradually increasing the “gain” on my Focusrite audio interface. There seemed to be a very fine line where when I turned it up too much – thus allowing for the RMS values to get more in line, then the noise floor (when speaking) would be too much. I could still do a “silent” recording and the noise floor was fine - with no error. So, I lowered my gain, got closer to the mic, and was successful at having the clip pass the ACX Check test.
I will take your advice on recording slightly to the side of the mic and see what happens. Also, help me understand something. How anal do it get with trying to get an error-free recording “raw” vs. getting close and then applying the Equalizer, Normalizer, and Noise Reduction (to apply room tone) effects? Oh, and I did download and install the LF_rolloff… equalizer plugin(?) you suggested. I seem to get conflicting opinions – having watched some of the ACX videos, they recommend staying away from some of the “effects” (i.e., compressors, etc.) – but then so many of the people who “do this kind of recording” appear to use the effects quite a bit. So, I’m a little confused as to what to really use and not to use.
Sorry for all the words and questions . . . but I always figure the only stupid question is the one you don’t ask (if you don’t know the answer - smile). Any additional recommendations you can give me will be hugely appreciated. Thanks so much again for your help and hope you have a terrific day!
“Audacity won’t record any more. It worked for months.”
And that’s the whole post.
some of it was a bit over my head.
We go until somebody stops us.
How anal do it get with trying to get an error-free recording “raw”
Any corrections you make in post production you have to do to every chapter in every book, forever. Doesn’t that sound appealing? Just be aware of that.
recording slightly to the side of the mic
Close. The mic is still pointed to you, just at your cheek instead of your lips. Move the microphone two or three inches to the left and turn it slightly. Directly in front of your lips is where the plosive pop sounds live. That also frees up the space directly in front of your face so you can read the script.
get conflicting opinions
We’re using the same words and mean different things.
We mean gracefully and gently applying minimal corrections to your already almost perfect recording to produce an ACX-Compliant work that sounds exactly like you.
They mean don’t even think about recording garbage and then trying to disaster recover it into a submittable story. Those have a high failure rate (unrecoverable) and are distressingly common. They also take up enormous ACX resources because ACX tries to tell the poster what they did wrong instead of just saying, “OK, you passed. Next!”
It’s rough to get people to think about what’s going to happen to their career if they get popular or get a contract for paid work.
This has happened several times: Remember that post production where you go in and correct your reading word by word? Good luck with that. Here’s a contract to read three novels.
If you get a contract for advertising or other announcing, it’s entirely possible you may need to filter your voice to be more dense, forward or aggressive. That can’t be done starting with a noisy or ratty recording. Can’t do it. All of those tools make background noise worse.
There was a recent posting pointing to someone on YouTube who claimed to have special Audacity settings to make your recording more professional. And it did, assuming you were recording in a completely soundproofed studio with professional (real professional, read: expensive) microphones and mixers as he was. There was a graphic about half-way through where you could see the special soundproof wall conditioning and heavy drapes he had.
If you didn’t start out with an already perfect recording, you failed.
At the end of the day, the YouTube posting was an advertisement for his services.
Thanks so much for sharing all the information and a giggle or two. I adjusted my mic yesterday and had good success – even removed my pop filter. But I’m going to adjust it again with the additional details/clarification you’ve given me. I’m also going to beef up some more soundproofing directly around my mic. I seem to be OK with not having the noise floor limits exceeded very often, but . . . I figure if I add some more sound absorption, it can only create a more consistent environment, right?
I discovered yesterday that I had more trouble with my peak limits exceeding the required range. I was recording more character-related dialog from a novel. So, it may have been the varying inflection and emotion in my voice. I’ll see how I fair after making your recommended adjustments to the mic.
Koz, your help and expertise – and willingness to take the time to share it and help all of us – is truly AMAZING – and HUGELY appreciated!! I’ll let you know how the new mic adjustment and added soundproofing work out. Maybe I’ll even record a new sample and send so you can see if it’s better. Hey, take good care and have a super rest of your day!!
It’s not unusual for me to be introduced as “The one with the emails.”
Painting with verbs.
I’m going to adjust it again with the additional details/clarification you’ve given me.
Listen to it.
A friend of mine records in a completely odd way I would not have guessed based on my training. He is delighted with the work…and so is the client.
it can only create a more consistent environment, right?
Wear your Producer (upper case intentional) hat for a second. Would you hire someone who can do a reliable, high quality, one-pass recording, or someone who produces sub-standard work that might require hours of correction?
I had more trouble with my peak limits exceeding the required range.
Oddly, that’s a different tool. Effect > Limiter. Compressor tries to reduce the volume of the whole work on the fly to make it appear more dense and loud. Limiter tries to affect only sounds louder than a set volume. It’s perfect for taming blue wave peaks when the rest of the performance is perfect.
Thanks again for your input. I have not had a chance to seriously try out my new setup yet – I changed several things to hopefully make my recording space more comfortable and to increase consistency in my recording results. Guess I trying to be optimistic that I may actually get my foot in the door and get something serious someday. Anyway, I’m curious about your reference to your friend who records in an odd way. You have Listen to it above that statement – so, I wasn’t sure if you had intended that to be a link to a sound bite. Just curious as to what he does that is different than most?
OK, going to try and get upstairs now and play with my new setup and see what may still need tweaking. I do notice sometimes that I can record the “silence” and not get a Noise Floor error, but sometimes when I physically record something, then that error pops up in the ACX Check results. Any thoughts on what makes that happen? Hey, hope you have a great rest of your day and a super weekend! Take good care!
I was commissioned to shoot the Los Angeles half of a broadcast radio program, so my microphone is on the left and we double recorded it.
You can just barely see it, but his microphone is not aimed toward the performer’s chair. Someone figured out that you could get tonal and timber voice variations with this microphone by not aiming it correctly. So he routinely records his material with the microphone pointed toward the credenza rather than his mouth [cringe]. Because he also has that semi-circular sound baffle and good office sound-proofing, it works. He got lucky.
a link to a sound bite.
That’s the unfortunate appearance of using underscore as theatrical emphasis rather than a link.
If you have a reasonable sound system or good headphones, stop what you’re doing every so often and listen to the work. It should sound natural, it should sound like you and there should not be anything obvious left when you stop talking. It’s fun to dredge in exotic and hard to use filters and tools, but you should have realistic goals and the goal should not be “fun with filters.” I have been known to sleep on it and take it again fresh the next day.
You should fix failures such as the peak error thing. That one is real and it’s good to get a feel for how all the tools work.
If you want to be obsessive, you can exceed the ACX specifications (just not by too much). Careful reading is good.
measure between -23dB and -18dB RMS and have -3dB peak values and a maximum -60dB noise floor
Their words, but that’s actually not true. The ideal RMS (loudness) really is half-way between -23 and -18, and the maximum noise is -60, but the ideal peak value is less than -3. If you try to hit the peak value exactly it will make you crazy, particularly because conversion to ACX MP3 submission format can affect peak values.
Try that once. Create a “chapter,” process it, export as ACX standard MP3.
Scroll down to submission tricks.
Open the MP3 in a fresh Audacity and see what happened to ACX Test.
MP3 is a terrible production format. The volume, quality and duration of MP3 files changes.
I think ACX Test will complain if it can’t find at least a half-second of room tone to test noise. It may also not like really long submissions. Does it say anything about that in the download dialog?
Me again. Well, maybe the moon and stars just weren’t lining up right for me to get a good recording today. Actually, I became a bit discouraged and here’s why:
When I record “room silence” – all is well – no Noise Floor issues. But, several times when I recorded, the Noise Floor was out of acceptable range. (Do I need soundproofing above the mic? Right now it is only behind me, to my left, and behind the mic.)
If I get close enough (at least that is what appears to affect it) to the mic to get my RMS level in range, then I usually get the Peak Levels out of range – and vice versa. My mic volume is maxed out at 1.0 and the gain knob on my Focusrite box is turned a good 3/4 the way around. Following are some of the various numbers in my “unacceptable” (raw recording - ACX Check test) recording results:
RMS results: -23.4, -23.5, -23.3, -23.1, -23.2 (You can see they are very close.)
Peak Levels results: -2.9, -2.8, -2.4 (Knew you said about using the Limiter – but . . . is ACX OK with using that?)
Also, maybe I’m still not grasping some of the things you are recommending to me. Am I trying to hard to always get a perfect “raw” recording? If so, I don’t think that’s going to happen. Just hope you can give me some recommendations on what to try – or encouragement to relax a little and be OK with some minimal “mastering/post-production” effects applied. Oh, and I suppose most importantly . . . and I do hope and trust that you will be brutally honest with me . . . even with the one little sample file I sent you, do you find my voice to even “sound” like something anyone would want to listen to? “No” is a perfectly acceptable answer, as I’m always up for brutal honesty.
Well . . . hey, it’s Friday and 5:00pm somewhere (almost here - smile) . . . so perhaps I’ll just step away from it, have an adult beverage, and figure that tomorrow is another day. Sorry again for all the words, but I hope I explained myself clearly. Enjoy your evening and weekend!
ACX isn’t OK with doing anything. If all you need is tiny corrections, we do it and don’t tell anybody. As I posted, the only people seriously burned are the performers producing actual damaged work.
But, several times when I recorded, the Noise Floor was out of acceptable range.
So it’s moving…? How far out? I don’t suppose you can tell what the noise is? I have a wall clock that can destroy recordings even though in real life, it’s not all that loud. There are some lamps I can’t turn on.
Undimmed old tungsten incandescent lamps are the quiet heros. Long-Tube Flourescent lamps are terrible, dimmed or not. Most of my compact flourescents make noise. Jury’s out on the LED replacement series. They appear quiet.
Fuzzy Generality Time.
ACX stresses quality continuity between the chapters. Also between books, but they’re particularly sensitive to chapters. It’s possible you may need a suite of corrections, no one tool enough to be obvious, but the whole suite enough to correct the average of all the chapter errors. Noise Reduction (for example) applied to all of them even though only chapters 3, 5, and 13-15 actually need it. The compression collection (Normalize-Compress-Normalize) should solve the RMS (loudness) problem without exceeding the peak specification. If you apply it to all the chapters it should not be obvious what you’re doing, but magically, the wimpy (low RMS) chapters vanish.
And this is where I go into the mud. Noise Reduction starts the process with a profile where you take a small sample of the noise by itself so Noise Reduction knows what to attack later. If your noise is moving, the sample and the chapter could be different.
I do hope and trust that you will be brutally honest with me
Voice and presenting style are between you and the client. In general, the only time I get involved is if your natural presenting style is impossible to record. And even then…
Once the recorder is working, you join the other actors looking for work. What does the Producer/Director/Client want?
Volunteer. Several participants got much good experience by volunteering voice/performance announcing.
Check with the Public Library. Ever read for a room full of kids?
There’s no One Size Fitz All correction, but since you are wandering in and out of compliance with no help, you might try these. The first one is “noise reduction of the beast.” If you get a good, clean noise profile at the beginning of the reading, you should be good to go without getting a fresh one for all the other chapters. I admit to being worried about your background noise changing over time…
Noise Reduction
– Drag-select Room Tone, silence or the flat area between spoken phrases.
– Effect > Noise Reduction: Profile
– Select the whole clip or show by clicking just above MUTE.
– Effect > Noise Reduction: Settings 6, 6, 6 > OK
Followed by gentle compression (slight recommendation change from earlier).
Audio Compressor
– Select the whole clip or show by clicking just above MUTE.
– Effect > Normalize: [X]Remove DC, [X]Normalize to -3.5 > OK
– Effect > Compressor: Thresh -20, Floor -40, Ratio 2:1, Attack 0.2, Release 1.0, > OK
– Effect > Normalize: [X]Remove DC, [X]Normalize to -3.5 > OK
Test with ACX-Check. All the chapters should pass whether they needed “help” or not, and the chapters should more or less match.
Those settings require you be close to compliance at the beginning. If you’re not, then you’ll need more Disaster Recovery tools. That’s one reason I haven’t published before now.
Understand. Thanks so much for the added details and recommendations. I also understand the need to have the raw recordings as close to “accepted” as possible. Since the last time I wrote you, I back my gain down a bit on my Focusrite box. That appears to have solved both my noise floor errors and my peak levels being out of range. All that has been left is more than not, the RMS levels slightly “unacceptable”.
I had been aware of the Noise Reduction effect and how to use it, as well as the normalize and compressor too. I had watched a video done by the audio geek (smile) for ACX . . . and he provided a list of the 3 effects in the mastering process for which they deem acceptable. Plus, I had been to your website and found various settings and recommendations for some of the ones you listed above. I just completed making an MP3 file so I will do what you had recommended about bringing it back into Audacity and running the test. I do make sure on any of the raw files, that using a few effects as possible that the recording passes.
WOULD REALLY APPRECIATE YOUR OPINION ON THE ITEM BELOW:
Lastly, and I maybe should have made this a new thread. I recognize that my “S’s” are quite sharp and hissy. I downloaded the “De-Esser” (spell?) plugin that Trebor used on my initial raw submission file. It does work quite nicely. Is it acceptable (to ACX) to use that? I also downloaded the De-Clicker he used and recommended. And, I have to say, it does a nice job as well. Just wondering how you feel about using either/both of those – especially the De-Esser one.
Again, thank you for the list of effects and the settings. I really do hope I can make this work, as I think doing this kind of work would really be enjoyable. I’ve always been an avid reader and remember in first grade (yeah, I can even remember back that far - ha! ha!) how I used to LOVE reading out loud in our “Weekly Reader” group. So, we’ll see.
Hope you have a great rest of your weekend. Again, I sincerely appreciate your help! Take good care!
Of course not. ACX doesn’t like us using any post production filters and tools. Noise Reduction and De-Esser are specifically listed in their forbidden graph.
There’s a cartoon with a monkey pulling an ostrich head out of the sand ignoring the rest of the bird (also sticking out of the sand) standing behind him. The caption works out to “Everything is connected to everything else.”
The aggressive, sharp, harsh sound in some microphones is their effort to be “professional.” Those sharp, punchy SSSibilants are a major contributor to you missing RMS/Peak compliance. You may easily find that after you de-sibilantize your work (to coin a phrase), you don’t need the compressor step any more to make compliance. I use older dynamic (moving coil) microphones that don’t have that sharp, crisp sound and I went straight to compliance. The closest I came to failing was noise, not peaks.
I think it’s knee-slappingly funny that manufacturers make microphones bright like this and then the performing public made a significant industry in software to take the brightness out.
Afraid that was the answer. I had seen the “acceptable” things that ACX mentioned in the “Mastering” video I watched (what you posted in your last reply). Again, though, I read so many places where people apparently think nothing of using multiple effect filters. Like Trebo replying about having applied the De-Clicker to my raw file. When we get replies like that, it is hard to decipher what is allowable by ACX vs. what maybe works in post processing for personal use? I truly DO desire to get a great “raw” recording. OK, so . . . what microphone do you use or recommend? Since my last post, I have done some research on the hissing S and have found some potential things to try . . . such as pointing the mic a bit down and away from my mouth (approx. 10-15 degrees) and actually being considerably back further (12-18") from it than I have been. I’ll try that . . . but your point is a good one, with regard to what mic you use. So, please reveal.
Again from up the post, there’s a big difference between someone in the “Mastering” phase of their work to produce a client deliverable from a perfect studio recording, and someone doing Disaster Recovery. ACX gets a lot of disasters.
I know where they’re coming from. If you can’t produce a good, almost perfect recording without the crutches, we don’t want to hear about it (but it’s bad form to say so).
Sooner or later someone was going to ask me that and I don’t have a ready answer. I’m coming from broadcast and I certainly have opinions, but not necessarily recommendations. I would have no trouble making a recording with an ElectroVoice RE20. Lovely microphone somewhere in the $500 range. One of my test recordings was a Beyer M58 field microphone somewhere in the $300 range. The M58 plugged into an actual sound mixer, my Peavey PM6, not a “USB adapter device.” That’s why I was so excited when the Behringer UM2 showed up. It was a USB microphone adapter that wasn’t terrible. I would still be leery about shooting a show with one, but there it is.
I have a Shure FP33 field sound mixer, or most of one.
I sweet-talked a friend of mine in Florida to give me his old one when he bought a new one. That’s what it looks like after you shoot sound in a speeding boat on Miami’s Biscayne Bay. One of the two channels still works. Those go for around $1500. Without the USB adapter.
The microphone in the voice capture pictures…
…is a Rock Band SM-58 microphone working with a Shure FP24 mixer (not an FP33). I would not do that again. The combined performance is not up to full quality standards. It’s too electrically noisy and I had volume problems.
This isn’t dreadful.
That’s an ES58 with my mixer and USB adapter. The original microphone SM58 sounds OK but has low volume. The ES58 is louder, but doesn’t sound as good.
Isn’t this fun?
So no, I don’t have a convenient laundry list of equipment. Most people recoil at the idea of learning how to use a sound mixer…
…but I wouldn’t do it any other way. Not necessarily vastly better than other methods, but for one example, it’s nice to know I have three different methods of adjusting the microphone volume. All my USB adapters have one volume control and it’s almost always used full up.
I adjusted my mic yesterday and had good success – even removed my pop filter. But I’m going to adjust it again with the additional details/clarification you’ve given me. I’m also going to beef up some more soundproofing directly around my mic. I seem to be OK with not having the noise floor limits exceeded very often, but . . . I figure if I add some more sound absorption, it can only create a more consistent environment, right?
