Onboarding Guide for Narrafix Proofeditors

This guide introduces the Narrafix Proofediting Workflow and helps users quickly become productive when working on audiobook productions with Narrafix Studio.

It is designed for:

  • proofreaders who are new to audio editing
  • audio editors who are new to script proofing
  • freelancers onboarding into Narrafix-based proofediting workflows

The goal of this guide is to explain how proof editors can efficiently review and finalize audiobook recordings with Narrafix.

1. Understanding the Narrafix Proofediting Workflow

Traditionally audiobook production involves multiple separate steps usually performed by different specialists. The different modules inside Narrafix does help significantly reduce the workload of production staff in many ways:


Narrafix automatically:

✔removes retakes

✔removes duplicate sentences

✔ aligns speech with the script

✔ detects potential script deviations

✔generates proofing markers


Narrafix automates large parts of the editing process and checks recordings against the script. But even though Narrafix does 90-95% of the work, it does not deliver distribution-ready files: There are still edge cases, pronunciation details and other things that need a skilled ear and human judgement for a quality product. Your work therefore focuses on quality control and fine adjustments. Your role is to perform the final human review and refinement.


Your job will be:

✔ listen through the audiobook once

✔ verify and correct automated edits

✔ check the recording against the script

✔ review pronunciation

✔ clean up and review/adjust pacing, breaths and noises

✔ flag corrections for pickups or remove unnecessary auto-flagged corrections


This means the proof editor performs ✂️editing and ✅ proofing simultaneously in one pass.

⚠️ Important: This is the only stage where the audiobook is listened to in full detail before delivery. Any error missed here will likely reach distribution.

Typical working speed

Onboarded proof editors usually work at approximately 1.5× the audiobook runtime

Example:

Audiobook runtime Typical proof-editing time
1 hour ~1.5 hours
5 hours ~7.5 hours

If you consistently work significantly slower or faster, discuss this with the production team.


2. Getting Started in Narrafix Studio

Before continuing, make sure you are familiar with Narrafix Studio, the browser-based editor used for reviewing and refining the results of the Auto-Cut and Proofing modules in Narrafix. If you have not used Narrafix Studio before, please start with this guide:


➡ Getting Started with Narrafix Studio


Once you understand the basic layout and navigation of Narrafix Studio, return to this guide to learn how to perform proof editing efficiently within the Narrafix workflow.


3. What to deal with in proofediting

There are several things to watch out for during the proof editing phase. It is essential that you stay focused and alert throughout, because your role is to catch the important details that studios, narrators, and algorithms may have missed — the details that make the difference between an acceptable result and a high-quality audiobook listeners will truly enjoy.

Always make sure you are working with full concentration. If you notice your attention fading, take a break and return once you feel rested and focused again.

In the following sections, you will learn which areas require particular attention, which kinds of issues commonly occur, and how to handle them. The goal is to help you develop a clear sense of when to intervene and when not to. Please read the following guidance carefully and in full.


3.1 Misreads and Errors

Correction markers generated by Narrafix are already embedded in the WAV files. However, remember: Always maintain full concentration while listening!


Narrafix:

❌ does not check pronunciations

❌  may not detect every error


Rules: If a marker is meaningless, simply delete it. If a marker is correct, leave it there and give more input/context if needed.


Missing Content

If you notice that more than one page or an entire chapter is missing, immediately notify production and set a clearly labeled marker. If large sections are discovered only during the correction phase, deadlines may be missed. If only a short passage or sentence is missing, creating a correction marker is sufficient.

Handling Editing Errors by the Software

If Narrafix fails to detect duplicate takes, manually delete the incorrect take (be careful not to cut too much!)

If passages are cut off or missing entirely, you can restore it easily by expanding the cut in Narrafix Studio. If there is no cut marker shown, you can be sure that the missing bit does not exist in the uploaded material. Place a marker there, if not already auto-generated.

Notes on Script Errors

Voice talents are instructed to read exactly what is written in the script, even if the script appears incorrect.  Therefore: Do not mark passages as incorrect if you are not 100% certain.

Narrators sometimes correct obvious errors automatically. If a narrator has unintentionally changed the text or if there is uncertainty about the script: Create a correction marker and request a new recording.


3.2 Pronunciation

Ensure that proper names are pronounced consistently throughout the entire audiobook.

This is especially important in productions with multiple voice talents.

  • Example: One narrator pronounces a name in English while another pronounces it in German.

This must not happen. If publishers provide pronunciation guides (like SmartScripts or external documents), these must be followed exactly. If you believe a pronunciation guide is incorrect, notify the production staff so the issue can be clarified with the publisher.




3.3 Reviewing and Adjusting Gap and Pause Lengths

Pauses and gaps are important dramatic and structural elements. It is essential to ensure that pauses make sense and support a good listening flow. Many narrators already handle this very well, Narrafix Auto-Cut deals fine with them also, but sometimes issues occur as a relic of automated rough editing or due to limited audiobook experience, especially with new narrators. These issues often appear as pauses that are too long or too short.


Chapter Start and Chapter End Pauses

Narrafix usually deals with this automatically, but you should check back on this. At the beginning of each chapter/file:

0.5 seconds pause

At the end of each file:

2 seconds pause

This ensures that:

• chapters start almost immediately on streaming platforms

• transitions between chapters are clearly distinguishable when listening


Intro / Outro Transitions

Between:

• the title announcement (intro) and the book text

• the end of the book text and the outro

there should be a clear pause that fits the narrator's speaking tempo.

This ensures that the announcements do not blend into the running text and listeners recognize them as intro or outro sections. Benchmark: approx. 2 seconds pause


Adjust Speech Flow

Like mentioned above, some audiobooks have gaps or eve an overall speech flow that feels too fast or too slow. Make sure pauses between sentences are not too short or too long. (Using Streamdeck pause commands or 0,2 and 0,5 short command buttons in Narrafix Studio is the easiest way to adjust these.)

Many narrators briefly read ahead in their mind before speaking a sentence or paragraph. If such pauses appear at illogical positions (e.g., in the middle of a sentence or between closely connected sentences), they disrupt the reading flow and must be corrected.


3.4 Handling Noise and Breaths

After Mastering, Auto-Cut and MagiClean (if applied), the most of noises and breaths should be fine and relatively clean. However, during proofediting, all significant noises and distracting breaths should be reviewed and adjusted if needed The following sections provide an overview of typical cases.

⚠️ Important Note

You will need to develop a sense of balance between being too strict and letting too much pass. There is always a risk of:

  • editing too aggressively
  • leaving too many issues untreated

If you are unsure, ask the production staff and when in doubt, mark too much rather than too little.

Also keep the benchmark editing speed in mind: A routine engineer usually needs about 1.5× the audiobook runtime for proofediting. If you consistently take significantly longer, you may be editing too much or there may be issues with the material.


3.2.1 Noises

Below are typical types of noise that should be addressed.


Note on Audio Quality / Mastering

In principle, the audio should already sound correct and polished at this stage, since mastering adjustments may already have been applied.

However, during some productions it may happen that  recording setups change, certain chapters sound different, clipping, mastering or compression artifacts occur. If you notice anything disturbing or unusual in general that exceeds your job and capabilities to fix, report it immediately to the production staff before continuing proofediting.


Mouth Noises

Examples:

• lip smacks

• swallowing sounds

• mouth clicks

Many clicks are already minimized during mastering. Often they can be removed by cutting the micro-pause between letters, leaving only the part where speech occurs. This allows the speech to mask the remaining noise. If mouth noises occur during pauses, they can usually be selected and deleted easily.


Plosive Pops

These noises occur frequently. They can often be fixed simply by deleting the short affected section. Be careful not to remove too much surrounding audio.


Clothing Rustle

If it occurs during speech and is distracting/too loud

→ create a correction marker for the narrator

If it occurs during pauses

→ delete it

If this happens frequently, consider giving feedback to narrator or production staff.


Clipping / Distortion

If the narrator speaks too loudly and his setup is not adjusted accordingly, the audio may clip and sound distorted.

In this case:

• set an comment marker and label it as clipping


Noise / Hum

If noise or humming appears only in certain sections:

→ place internal markers at those positions.

If it occurs consistently throughout an entire chapter:

→ place one marker at the chapter start and inform production staff.

If noise occurs audibly together with speech, a correction marker for the narrator is required.



3.2.2 Breaths

Breathing is a natural part of speech and should normally remain untouched. Breaths should never be removed completely by default, unless there is a strong reason (e.g., noise within the breath). Below are cases where editing may be necessary.


Double Breathing

Two consecutive inhalations before a sentence sound unnatural. This can be a relic of automated editing or narrator-side editing. Delete one of the breaths, but make sure:

• the end syllable of the previous sentence is not cut

• the beginning of the next sentence remains intact.


Cut-Off Breaths

Sometimes breaths sound abruptly truncated. Possible causes are automated editing or narrator rough editing.

Solutions:

• smooth with fade-in / fade-out

• if necessary, remove the breath


Abrupt Breathing

Some narrators inhale very loudly or abruptly. This should be handled by the MagiClean-Module. But if not, these breaths can usually be fixed by reducing their level by –1 dB (or –6 dB in extreme cases) To avoid sudden entry, add a fade-in at the beginning.

If a narrator consistently has unnatural breath cutoffs or unclear word endings, it may indicate a gate setting problem. In that case contact the production team.


Continuous Breathing

Sometimes a low-level breath can be heard continuously between words or sentences. These breaths can usually remain. However, when shortening pauses ensure that no obvious edit cut becomes audible due to the breath sound.

One method is:

• remove half of the breath and fade the remaining part in or out.


Breaths with Mouth Clicks

Breaths sometimes begin with sharp transients or clicks. These transients can be deleted. Afterwards, ensure that the breath begins with a fade-in so it does not enter abruptly. Use the Delete key only if doing so does not cause the breath to start unnaturally early. Otherwise, use the Silence function (Backspace).


⚠️ Important Reminder

This is the only stage where the audiobook is listened to in full detail before delivery. Any error missed here will likely reach distribution.

Did this answer your question? Thanks for the feedback There was a problem submitting your feedback. Please try again later.

Still need help? Contact Us Contact Us