Feedback Request on Sample Recording for Audiobook

Hello - Hoping for tips from this forum’s expert ears. My voice isn’t at its best after consuming tomato sauce, but hopefully it’s enough to say if the room tone, levels, etc would be alright before I continue. Used a Zoom H1n, gain knob around 5, no other presets turned on.

The background sounded a bit buzzy to me with headphones, but maybe the audiobook mastering effects would take care of it?

I’m new to the H1n, so possibly also some user error there. I had it set to record in 96kHz/24-bit and then moved the file into Audacity from the microSD card to export as WAV (Microsoft) signed 16-bit PCM. Recorded in a small closet with blankets and pillows surrounding all surfaces with a laptop (no fan sounds from it) to read from and a lamp that seemed to be silent to me.

Thank you for any feedback!

First observations: My H1n volume control, that little rotary knob up near the microphones, seems to be happy at 9. The sample is very low volume.

To give you a better idea of your job, you’re playing three people. The Producer, the Recording Engineer, and the Performer/Artist.

The Producer makes the global decisions (which platform should we shoot for and submit to), decides whether or not to end sentences with prepositions, signs the checks, and hires the other two.

The Recording Engineer knows about soundproofing a room and how to drive the microphone and recorder. They define performer actions, position, spacing, and volumes.

The Artist/Performer knows how to interpret the work and read it in an entertaining, pleasant, and accurate manner. They show up to the studio just before the recording, do the work, and go home.

I applied Audiobook Mastering followed by a slightly stiff Noise Reduction (9, 6, 6) and the submission passes ACX standards.

Screen Shot 2024-03-16 at 9.26.20 AM

Even though the sample isn’t paid work, that doesn’t mean we’re not paying attention. The words you used in the sample aren’t as written, and there are other problems.

As we go.

Koz

If I didn’t say so earlier, you have a very pleasant recording voice.

I had to look up a couple of settings. I haven’t adjusted them in a dog’s age.

The four buttons just below the screen change every time you press them. I don’t know that the manual says that anywhere.

– Audio: Press it until the indicator reads 44.1/16-bit. Your sample is at some other sound standard.

– Lo Cut: That’s the rumble filter. Your woman’s voice should work well with 120Hz

– Limiter: Off. Jury’s out on this one. It could be handy depending on how they implemented it.

– AutoLevel: Off

As mentioned, my volume control up near the microphones is set to 9. They recommend occasional peaks on the bouncing sound meters should settle at -12dB.

The top of the recorder should be aimed at the performer. If you’re worried about popping your P’s push the recorder to one side and aim at your cheek/corner of your mouth.

There is one ACX recommendation that can cause problems. They want you to submit a mono voice track. One Blue Wave.

Mono_2024-03-16

They don’t demand this, but it’s highly recommended. If you do submit stereo (two blue waves), then your whole book has to match.

The H1n naturally produces stereo.

Select (far left of the track) > Tracks > Mix > Mix stereo down to mono.

Koz

Ah, good to know. I originally had it turned up higher but got worried I might be overdoing it with the sound meter. Will try again with it set higher

Guilty, as charged. I moved a word in to give myself another enunciation challenge

Thank you for looking those up. Will give this a try. I had the audio at 96kHz/24-bit thinking it would be better, but will knock that down.

I had it sitting at around 10 o’clock. Stereo-wise I definitely caught more in my left ear when playing it back. I intend to submit in mono; however, I have found one platform that requires stereo if you have any file less than 3 mins (which I will). In that case, would the best order of operation be: record in stereo → edit in mono (applying mix stereo down to mono) → apply audiobook mastering → save in mono for ACX → save mono version as stereo?

Thank you, Koz! I appreciate the encouragement. And the feedback for all of the ways I can make the audio better. Have learned a lot.

ACX Requires 44.1/16 submission. That’s Audio CD Standard if you’re counting.

https://help.acx.com/s/article/acx-audio-submission-requirements

`
Everything is better in mono.

You will find that the noise (ffffff) contributed by the electronics goes down significantly when you do the mix-down. In the case of your first test, it almost gets you past the noise requirement. It made it possible to apply gentle Noise Reduction of the Beast (6, 6, 6) and pass.

All of the tools, patches, and effects in the audiobook arsenal were chosen to be effective, but unobtrusive. ACX can’t tell what you did.

Editing is much easier and faster in Mono. If you’re looking for that one lip smack, searching one blue wave is much handier than having to juggle two.

Mono takes up half the storage space.

Mono transmits twice as fast and more efficiently than Stereo.

ACX recommends Mono submissions.

If you have a client that really, really wants stereo, you can mono edit and do post-production processing and then “convert” to “stereo” later.

–Select the whole performance with the button far left of the time line.
–Edit > Duplicate. I had to look that one up. I use the keyboard shortcut, Command+D.
–There’s a little down arrow (sample) to the left of the track. > Make Stereo Track.

I know you’re going to say: “That’s not real stereo. My voice doesn’t wander left to right as I speak.” Correct. but voices are special purpose. In a theatrical movie when you enjoy the environment surround sounds and the left-to-right effects and production tricks, the performer voices come out of the one single center track.

Movie Magic.

My To-Do list contains an abbreviated “reading an audiobook step-by-step.” It’s remarkably difficult. How many days have we been at this?

One thing we didn’t cover is the two Safety Backups. When you get done reading a chapter, Export it as a WAV backup, errors, fluffs, and all. This is less important for you because you have the WAV hardware backup on the H1n. This is against Audacity going into the dirt while you’re editing and taking the show with it.

When you’re finished editing and ready to make the submission MP3, File > Export a WAV Edit Master first. Once you make an MP3, you can’t edit or change it.

Do submit a new voice test to the forum.

Koz

Oh. One more. I’m going to experiment with the H1n Limiter.

As we go.

Koz

Going over notes:

Errors.

You may find your own technique, but this works for me.

When I know I made a speaking mistake, leave the recorder running, look back in the script to the next even sentence and read it all again. This gives a correction with matching pitch, emphasis, and rhythm as the original.

DO NOT depend on going back a half-hour later and only then reading corrections.

Some performers ring a bell or clap to make the correction easier to find. You’re intentionally creating a celebrity blue wave.

Koz

Thank you again for all of the tips. Will follow your advice.

New samples here. H1n settings: 44.1/16-bit, 120Hz, off, off, gain knob at 9. It’s the same audio in the samples below, but when I import into Audacity the file is at 32-bit float. #2 is where I’ve left it at 32-bit float and exported as 16-bit. #3 is where I formatted it into 16-bit (selecting the down arrow on the audio file name → format) first and then exported as 16-bit. In case it makes a difference.


Left the mouth click in at the beginning (proof it’s unedited), but would have deleted it during editing.

I can only imagine. Hopefully this has added to the FAQ topics.

Will do. From an editing perspective, my current thinking is to edit and apply the audiobook mastering effects to all of the clips in one file (so all chapters sound the same) and then export chapters. The total recording time will be roughly 1 hour, so it’s not quite as large as most audiobooks, but will have 50+ parts.

Will stay tuned!

I think it’s down to which platforms libraries work with and unfortunately ACX doesn’t cover that. Overdrive seems to be where libraries go. I reached out to one of the ways to get to Overdrive (PublishDrive) and they told me their updated guidelines recommend stereo as “short mono files often cause issues during the pre-processing of the files”. Not sure if it’ll be worth the extra overhead, but good to know Audacity provides an option to convert to stereo later.

32-Float is another Production Note.

It’s possible to get standard, 16-bit digitized sound so loud that it “runs out of numbers” and stops following the show. That’s clipping (on the right in the illustration).

Clipping is Not Desirable. it sounds terrible—harsh, piercing, crunchy sound— and the damage is permanent.

Audacity works internally at 32-floating because effectively, it doesn’t overload or clip. This is super valuable in Effects. Audiobook Mastering, after applying a rumble filter, applies an RMS Loudness effect. This sets overall volume, but may easily push the tips and peaks of your blue waves into the overload region. But then Mastering runs a graceful, gentle peak limiter to ooze the peaks down and into conformance. This little dance only works in 32-floating.

If we did it right, you may not be able to tell anything happened by listening, but magically, ACX Peak and RMS now pass.

It’s possible to File > Export a 32-bit file, as long as you never expect clients to actually play it. It’s non-standard. So you have to convert it back to 16-bit for common use.

Koz

We tell people if you can pass ACX, you can usually submit your work anywhere else. At worst you may have to gently bump the volume a bit.

Koz

Sample2

You picked up some harshness in the “ss” tones. It’s not a clear voice any more, it’s a little sharp. Also, the background noise actually got worse.

You didn’t apply anything other than the mix-down to mono, right?

Try another test. Get rid of the LO CUT setting on the H1n and post the test in native stereo. Don’t mix anything.

Let Audacity do its 32-float thing.

Koz

I know this doesn’t help. but you’re giving me some ideas how to revise the test submission. Some of this text is a holdover from the old forum. It’s not needed any more.

Koz

That’s right. Just that.

Two fresh samples. One with Lo Cut and one without (in case there was something with the mix-down to mono before). Used ‘import’ in Audacity → cut out the parts where I hit the record button at the tail ends → exported.

With Lo Cut:

Without Lo Cut:

Is the background noise worse related to the gain being dialed up?

Always happy to inspire a process improvement! Can give it a test if helpful too.

These are the new instructions.

https://www.kozco.com/tech/audacity/TestClip/Record_A_Clip.html

Reload the page.

It mostly has to do with timing. The old forum would only accept sound tests up to 2MB. A slow reader of Catskill Cows might come in at 2.2MB and the forum would reject it [insert cursing here].

The new forum will allow up to 4MB tests. So we’re good.

There’s an Easter Egg in there as well. The Catskill Valley may be notable for a lot of things like fancy-pants cliff-side New York homes and West Point, but maybe not cows.

The cows are on the other side of the Catskill Mountains where I used to live. Most of my neighbors were cow persons.

Koz

Thanks for explaining that. I hadn’t touched the setting before the comment about the sample being at the wrong standard. I see that one wasn’t the troublemaker. Will continue leaving it alone so the effects can work their magic later

Looks good to me. I had skipped the front part of the text because I wanted to make sure I could get a ‘p’ in and include the ‘contented’ state of the cow persons. Knowing there’s more time/space is a helpful update.

I got With Low Cut to pass and sound good.

Now I need to go back and write down how I did it.

I’ve never heard a sound test do that before.

As we go.

Koz

Did your H1n come with two or three letters after the model number? I see there are a number of variations on this recorder.

There is Common Wisdom that all new microphones have to sound crisp and “professional.” They make my ears bleed.

As we go.

Koz