Greetings all, and thanks for sharing your expertise.
I’m a very new noob, trying to figure out if I’m under- or over-thinking audio quality. I’d greatly appreciate it if any of you more experienced ears could listen to a couple of cuts of audio and let me know how to improve my audio, adjust it for my voice, etc. Or tell me if I need to quit overthinking.
I’m attaching a couple of short cuts. These are from my first audiobook, which I’m working on for ACX. One is raw .wav file and the other has been processed with the Audacity ACX plugins. I feel like the processed cut is coming out sounding a bit tinny or hollow, and I realized that I just don’t have enough depth of experience to know what I’m listening for. Any help and criticism is greatly appreciated, and please don’t pull punches. (No ego riding on this. )
Equipment is low-end, so please let me know if you believe that’s an issue as well.
I’m using a Blue Nessie mic, Windows 10, Audacity 2.3.2. My booth is an inside closet with some blankets hanging down. I’m a little concerned about noise floor, since I have the laptop fan is usually going. It’s quiet, but constant.
You have a very nice voice that’s pleasant to listen to and the content sounds interesting. I’m not much more experienced than you, so absolutely don’t take me for an expert. I had the same problems about laptop fans and I got rid of that. I bought a Zoom h2N that the actual expert here recommended (a stand-alone recorder). Later I bought myself a new laptop on eBay for a song that’s absolutely quiet, so I could theoretically go back to using it, along with a mic that would probably cost the same as the Zoom. I’m just glad I have the laptop in there with me because I want to be able to make revisions to my manuscript as I record, and I certainly don’t want to be printing out lots of noisy paper to read from. In any case, IMO, figure out how to get rid of that fan. In hindsight, I’d say buy a new, silent little laptop and a bit upgraded USB mic (maybe you already have a decent mic).
You’ve got a chance at putting together a nice audio book. Go for it. And be sure you submit a sample to Audible that passes before you record the whole thing. I’m looking at it like community theater. You do a lot of rehearsals for that, right? It’s the same thing here. Lots of repetition till you’re sure you’re doing the best you can.
You should know that the only clean, pure room or background noise in your piece is the last half second or so. All the rest of the pauses have you breathing or other noises. That’s why the recommended production for forum posting has a 2-second hold-your-breath-and-don’t-move at the beginning.
Yes, I know. One of these days I’m going to pull this stuff into one place.
Anyway, it’s enough to pump very gentle noise reduction (6, 6, 6) in there and pass with very handy specs.
Fuzzy rule: any noise better than about -65dB should be good enough. People who struggle to pass the -60dB noise spec at -60.1dB are doomed. There is an actual ACX response where they say, Yes, you pass noise now see if you can do a lot better.
So now it’s between you and ACX.
I don’t have an illustration for this yet, but MANY people read from a noise-free tablet. You could read from your cellphone. Maybe you can do that, but I can’t.
The fuzzy rule for a stand-alone recorder is two weeks. If you’re still struggling with the computer past about two weeks, give up. You see those two, three and four chapter postings on the forum? Those are the people struggling with Everything that can Go Wrong with a Home Computer and a USB Microphone.
This is a Patched Posting with an actual poster response and me filling in extra info for how to post a test.
Export a WAV of each raw reading and save it in a safe place. Edit a copy. Do it again for your ACX posting even though you need to post MP3 to them.
You can’t edit or change an MP3 without adding sound damage. MP3 is an end product. Your readings and archive master should be in WAV (Microsoft) 16-bit.
Anything else you do is up to you.
There’s a 10,000 foot view of backups, too. (What’s that in metric?) Be able to point to two different places that hold valuable work. Thumb drives work, external drives work, cloud storage works. But there should be two of them and be different. That’s the silly joke about pointing. My test production is on Lucille (Pointing to laptop) and backup drive (pointing to shelf).
I recently had to rescue some old documents from a computer whose main drive went into the mud. I pulled out the second hard drive and was able to rescue the work [mopping brow].
If you don’t have a backup storage solution, Google Drive can hold you over until you can buy a thumb drive or something. If you really like Google Drive as a backup, you can pay 1.99 a month for 100GB of storage (improved from the free 15GB) or 2.99 a month for 200GB of storage, which will last quite some time. Their most expensive option is 299.99 a month for 30TB of data, but that’s really some overkill not worth considering.
I’d agree. I like the sound of your voice (definitely better than I enjoy mine, and mine passes). I’d work on your noise floor some, though it’s hard to analyze without some good room noise samples in your file.
The reason being that it might be passing now, but it’s not passing by much. Your mic and Audacity are going to pick up everything, including noises we can’t hear very well, or at all. If an airplane goes overhead, or a gentle breeze comes through your room, heater turns on, etc., you might run the risk of failing.
I feel like the processed cut is coming out sounding a bit tinny or hollow
It might. That first step, the equalization, takes some of the sub-bass notes out of the production. Intentionally. Your recording system will have a very difficult time passing muster without that. Your microphone has serious sub-audible rumble, earthquake and thunder tones. They all count towards making noise worse.
The down side of that filter is what happens if your male voice happens to have tones that low. We could try a slightly gentler equalization filter. You can’t leave it out.
This is the same filter that Hollywood uses for field production. So you and John Wayne have similar sound capture.
Your mic and Audacity are going to pick up everything, including noises we can’t hear very well, or at all.
The Kitchen Table Studio is a lot better at echoes than refrigerator noises. Echoes have to leave your space, go through the blankets, bounce off the walls, and come back through the blankets. It effectively kills echoes.
ACX -60dB noise, in English, means your studio noises have to be 1000 times quieter than your voice. If you can tell your computer is on by listening, you’re going to have a bad time.
Good to go even with no noise reduction at all. I get noise at -65.5dB with straight mastering.
However there are two cautions. Your raw reading has blue wave tips over 50% (0.5). That’s the 6dB limit for raw capture. If you hit 100% (easy to do at that volume) that may cause permanent sound damage on those sounds. It’s not pleasant. Popping, clicking, ticking and crunch.
If you look at the bouncing gradient sound meter during those segments, you may see it popping well up into the red. Keep that meter in the yellow or orange and the blue wave tips at 50% or below.
The bouncing sound meter can switch colors. I get there with Control-Click the playback meter > Options > Meter Style. In general, Gradient is good for recording because it changes colors during the performance as a warning. RMS is good during playback because it gives you loudness and peaks in the same meter.
The other caution is silence. When I did the mastering trip, I cut off the last 2-1/2 seconds of silence. The ACX loudness specification is based on RMS, and RMS is based on an overall average of energy in the show. If you’re too theatrical and emotive, the overall RMS may sink and throw off the other two readings.
Other than that, I think it’s time to post.
Again, we’re only dealing with the hardware issues. You also have to pass the theater tests. Please post whatever they say.
Once again, thank you Koz. I have to run off to real life, but will post on any feedback I get.
One quick question:
Keep that meter in the yellow or orange and the blue wave tips at 50% or below.
Do you recommend that I turn down the gain or back off from the mic? Or are you just saying to practice reading and being aware of the meter? (I did drag it and stretch it out, courtesy of a handy tip in one of your other threads.)
Or are you just saying to practice reading and being aware of the meter?
I’m saying do practice runs while listening to yourself in your live action headphones to get a feel for what it sounds like when you’re performing in range.
Do you recommend that I turn down the gain or back off from the mic?
That’s a class right there. If you have a directional microphone, backing away will change the quality of your voice. Close-in sounds intimate and throaty. Further away sounds crisper and cleaner, but more subject to echoes. Turning the volume down will change the noise level. So if you like the way you sound, stay there and sneak the volume down slightly. This is not an End Of The World volume change. You’re slightly too loud and it’s making me nervous about being too close to overload (100%).
Can you see Audacity from your studio? announce tests watching Audacity while you speak and listen. Shoot for occasional peaks at -10dB to -6dB.
You can run Audacity in Monitor Mode where the meters work but you don’t make any actual recordings. Click on the recording meters > Start Monitoring.
As always, thanks for letting us all benefit from your experience! So yes, with my current setup I can watch the meter while recording. I’m going to do some trials tonight to fine tune performance and input - and then it’s on!
Feeling excited to get home and practice. I also feel like I hit a lucky streak to have a closet that’s really quiet.
I sent ACX a sample for their qc a couple of days ago, and will see what their feedback is. But seriously, this forum has given me a big boost in confidence. Capturing audio well is clearly a craft that should be tuned and improved over the course of years. But it feels a lot more doable today than last week.
And post it. We build up our advice dictionaries by listing what works and what doesn’t and how we fixed stuff in the past.
For one example, ACX Check is totally home-grown by Flynwill assembling and organizing existing tools. ACX had nothing to do with it past publishing detailed specifications. Over time we were able to compare submissions, results and comments to the point that if you pass it, you stand a terrific chance of making it past ACX’s technical inspection.
But not always. The two inspections, the Robot and the Human occasionally overlap. They can fail your noise not because you actually failed noise, but you got past noise by beating your noisy voice with a stick until it turned to trash. That’s one of the things we can inspect for before we turn you over to them.
“They will never accept that. Your voice sounds like a broken cellphone on a windy day.” as an example. They count that as a distraction. They hate distractions. Your presentation should sound like somebody sitting next to you telling you a fascinating story.
This is a theatrical performance and final judgement is theirs.
This is a theatrical performance and final judgement is theirs.
What I keep seeing in the forums, too, is that there is the robot check and the human check.
There’s actually a middle check too. The author you’re submitting files to. They’re going to listen to what you’ve sent too. They may not be trained in the specifications of what ACX wants, and they may be even more clueless than you are, but you’ve got to impress them before you get a chance to impress ACX, sending in a quick sample aside.
When I auditioned, I aimed to make the sample sound pleasant for myself before sending anything in. It took me several days to make what amounted to 3 minutes of audio, but it worked and I learned quite a bit. The robot is easy. People are definitely trickier.
I actually have a project called 'Starter file" I use when I start editing.
It’s just four tracks. The first, second, and third tracks are all just room tone, repeated (with the effect) a few hundred times. Each track has different lengths of room tone. One is what I might use between paragraphs, one between sentences, and one for commas and short pauses. The fourth is just an extra track with nothing in it that I can record into if I don’t like something I’ve already done and want to replace it.
Then I just import audio, bring in my raw wav, move it to the top, and start to edit. Whenever I need some room tone, I have preset amounts I can just toss in. I can also just pull out a small blip of it if necessary.
The first and last thing I do when editing is to take the large chunk of room noise and smack it right on the start of the file. I go for 1 second at the start and 5 seconds at the end, this way if I trim into it a little bit while I’m editing, I’m still golden.
The second posting has actual graphic captures of their room tone requirements. This is a picture, so the links don’t work.
Everybody gets Room Tone wrong. That’s the Room (or microphone system) performing, not you breathing, gasping, coughing, shuffling, smoothing, rearranging and clearing your throat. You can’t get up and walk away because that will affect the room acoustic signature and, among other things, screw up noise reduction.
Isn’t this fun?
Sometimes we can tell exactly what the poster is doing by the acoustic trash they’re leaving.
“OK, this is where they scooched their chair in a little.”
“This is where they yawned.”
“This is them nervously tapping their wooden table.”
“This is where they started texting their cousin in Cincinnati.”