A Critical Ear: The ACX Reference Sample Pack

Hi! This is Brendan from the ACX QA Team. I’m here today to introduce our Reference Sample Pack, a new tool we’ve developed to illustrate how your audiobook should—and should not—sound during the various stages of production.

This tool will help you spot problems in your audio and give you an idea of the audio quality your listeners will be expecting from productions on Audible. We’ve also included files that can be used to calibrate your Digital Audio Workstation (DAW) for recording. Mastering level specifications, especially RMS, can be difficult to understand via text alone. What better way to learn what kind of audio “passes” ACX QA than to have passable files at hand for you to refer to and test on your own setup?

Getting Started

To use the Reference Sample Pack, download the zip file onto your computer. Unzip this folder and you will find nine WAV files that can be loaded into your DAW of choice. We processed, and in some cases distorted, the same raw file for each example, then divided the samples into two categories: files that can be used as good production targets, and files containing issues you should try to avoid.

What’s Inside

Start with our PDF guide, which contains exact details on what you should listen for while playing.


The “Good Production” Files

File 1: A Raw, Unedited File (good-production_01_raw-recording.wav)

This file has a few issues that need to be resolved before it can pass QA, the mouse clicks and excessive spacing at the start of the file for example, but nothing you hear can’t be resolved during the editing and mastering stages.

File 2: An Edited File (good-production_02_edited-recording.wav)

This next file contains the same performance, edited properly.


Notice the edits made between “Step 1” and “Step 2.” We trimmed the spacing (circled in purple) at the top of the file to half a second, and removed the mouse clicks and deep intake breaths (circled in yellow), replacing them with clean room tone.

Learning proper editing techniques can take some time, but I’ve found that the Alex the Audio Scientist blog post on editing and spacing is a helpful starting point. I even use the same QC sheet referenced in Alex’s post when I work on my own projects.

File 3: An Edited Master, Pre-Encoding (good-production_03_edited-mastered-recording.wav)

Ever wonder what a file that meets our Audio Submission Requirements, with peaks around -3dB and RMS levels between -18 and -23dB RMS, would look like in your DAW? This is it! Observe how consistent the peaks in this file are, then check where this file peaks on your meters. If your file is too dynamic or sounds a little muddy, you may need to utilize mastering tools like those detailed in Alex the Audio Scientist’s Mastering Audiobooks blog post

The “Avoid” Files

The “Avoid” files contain common problems you should steer clear of during production. Included here are:

  • A file that has been recorded at levels that are too low,
  • A file that’s been heavily gated,
  • A recording processed with heavy noise reduction, and
  • Files with Peak or RMS levels that do not meet our requirements.

You should not use these files to calibrate your system for recording. Rather, train your ears to notice these sounds as you work on your own files, and use these examples to understand the most common issues you may run into during production.

Try It Yourself

The sample pack also includes the script used during the recording of the samples. If you are testing out your levels before you begin a new project and want to compare your recordings to the “target” files in this pack, we recommend you use this script, which can be found on the last page of the file “ACX—Sample Guide.pdf.” Record your read of the script and compare your noise floor and peak level to the “Step 2” file. The closer you can get to matching the samples, the more confident you can be that you will pass QA inspection later on in the process.

Final Thoughts

It can be easy to get caught up in post-production, using too many plugins or tools when trying to meet specifications, or trying to fix poorly recorded audio that is beyond repair. At ACX, we believe the best time to address audio issues is before they make their way onto your recording. Training your ears to know when problems are occurring will be far more beneficial than having the latest noise removal or EQ plugins will ever be. The better you get at listening to yourself, the better your productions will sound to others.

Did you find the QA Team’s Reference Sample Pack helpful? Tell us in the comments below.

13 responses to “A Critical Ear: The ACX Reference Sample Pack

  1. First of all, thank you SO much for taking this initiative. Reference materials like these are crucial to anyone seeking to improve their audio quality, and engineering knowledge and skills.

    I’d like to suggest adding to the “bad” examples recordings made in poorly treated spaces. In browsing the samples uploaded to ACX daily, I hear many that have been recorded in very live spaces, and I think do/don’t examples demonstrating the difference between an untreated and a well-treated space would be instructive.

    Again, this effort is so needed, and so appreciated. Thank you for you continuing support of the audiobook production community, and our shared efforts to raise the quality of ACX titles.


    – Paul Heitsch

  2. Thanks. Look forward to trying this out.

  3. A great resource to help improve the finished product. Thanks Guys!

  4. Very useful! I have been a voice sound producer and engineer since the late 70s, and it is hard sometimes trying to convey the importance of all of this.
    If I can add a note on dynamics. Some people I talk to look at compression as a way to solve uneven voice levels during recording – sections of quiet speech, a nice loud argument and so on.
    Those problems should be sorted out BEFORE compressing. In studios, we often did this during recording. We kept our hand on the mic fader at all times, watching the script carefully and would ride the levels gently (or sometimes quite dramatically) as we recorded.When it came to mastering and we applied a final limit and compressor, then it was being used to make the voice sound louder, not paper over the cracks. As soon as you use it to rescue a recording, then you are in trouble.
    Now, when recording oneself, you cannot ride the level. However, on a decent DAW like Cubase, Studio One, ProTools and so on, it is a simple matter to adjust the levels during the editing process to make sure they are where they should be. You could even add automation, riding the level throughout, but that is probably a bit too longwinded. 🙂
    So, get your levels right first, and then final mastering will be adding polish.
    Likewise with gating. In the studios, we NEVER used gates on voices. Not even subtly. Even if you adjust the range so they do not cut out completely, you still risk them softening the beginnings and ends of words and shortening the ends of breaths making them sound like gasps. Not nice. A good breath is a boon to the performance – don’t mess it up with a gate. If you have a noise floor issue, throw more mattresses and quilts at the problem, not electronics.
    Great little resource.

  5. This is a great resource! Thanks! I do have a question about RMS levels with the good-production_step-01_raw-recording.wav file. You say “The RMS of the narration is -22dB RMS”, but when I look at Waveform Statistics in iZotope RX6, the Total RMS Level is -19.30dB. Here’s a screenshot of the dialogue box – https://www.screencast.com/t/jbCXQxp7wkgJ. I want my recordings to be within ACX specs and I’m now worried I might be calibrating to different norms. Thanks for your help!

  6. The entire article and links should just be: Hire a professional, because these files A.) Don’t sound that great, and B.) Don’t give nearly the explanation for people to understand the tools that they are using. Not only that, it doesn’t even pass your specs, the head is less than 0.5 seconds and there are points in which the noise floor exceeds -60db

  7. Thank you! This is really helpful!

  8. Art, click on the Edit tab in Izotope and then on the Preferences at the bottom. Then in the Miscellaneous tab un-check the Calculate RMS using AES17 and you will get the correct value in the Waveform Statistics.

  9. As someone new to voice over I found this extremely helpful. I have been working like crazy trying to get the right space and equipment. I am not sure I heard that much of a difference in the dynamic file. I would love to speak with someone more in-depth and maybe get one of my samples evaluated. Is there anyone I can reach out to that might help me with this. I am happy to pay for the time. Thank you so much for creating such an amazing space for talent to learn and grow.

    • Hi Cathi, Please reach out to support@acx.com and ask about our sample review process. We’d be happy to take a listen to one of your recordings and advise how it compares to our audio submission requirements – SJ

  10. Where do I find the RMS and noise floor numbers? I’ve gone through hours of YouTube videos and still – where are they??? are they on the track somewhere or what? Thank You.

  11. I went through all this and still no answer to my question: Where do I find the “-23 –18 RMS and -80db Noise floor” in Audacity?? Where are these numbers located?

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s