Audio Description Pipeline: Fill Dialogue Gaps
An audio description pipeline that fits the gaps in your dialogue is a workflow that inserts narrated descriptions of visual information into the natural pauses between lines of speech, ensuring viewers who are blind or low vision can follow the action without missing dialogue. You build it by combining a script-writing step, a recording step, and a mixing step that places descriptions precisely into those gaps. This article explains the components, the timing rules, and the common pitfalls so you can assemble a pipeline that works for your content.
Why Audio Description Must Fit Dialogue Gaps
Audio description (AD) is an additional narration track that describes what is happening on screen: actions, settings, facial expressions, and text that appears. The core constraint is that the description must not overlap with dialogue or important sound effects. Viewers rely on the original soundtrack for the story, so the AD must be inserted into existing pauses. If there are no pauses, the describer must either condense the description or, in some cases, the editor must slightly extend a pause (if the license allows). The goal is a seamless experience where the description feels like a natural part of the soundtrack.
A pipeline is the sequence of tools and steps you use to produce that AD track. It typically includes: analyzing the source video, writing the script, recording the narration, and mixing it into a final deliverable. Each step has specific considerations to ensure the descriptions fit the gaps.
Core Components of an AD Pipeline
Every AD pipeline, whether for a short YouTube video or a full-length feature, has the same five stages:
- Source analysis: Watch the video and note every visual element that is not conveyed by the audio. Also note the timing of dialogue and sound effects.
- Script writing: Write concise descriptions that fit into the identified gaps. The script must include timecodes (start and end times) for each description.
- Voice recording: Record a narrator reading the script. The recording must be clean, with no background noise, and the pace must match the available gap.
- Mixing: Place the recorded descriptions into the timeline, adjusting levels and equalization so the narration sits comfortably under or between dialogue.
- Quality control: Play back the final mix to ensure no overlap and that the descriptions are understandable and accurate.
For a simple pipeline, you can use free tools like Audacity for recording and editing, and a video editor like DaVinci Resolve or even iMovie for mixing. For professional work, you might use specialized AD software that helps with timing, but the principles are the same.
How to Identify and Measure the Gaps
The first practical step is to map the dialogue. Open your video in an editor and look at the audio waveform. The gaps are the quiet spaces between speech. But not all quiet spaces are usable: you need a gap that is long enough for a description to be spoken clearly. A typical rule of thumb is that a narrator speaks about 150 words per minute, or about 2.5 words per second. So a 2-second gap can hold roughly 5 words. But you also need a little breathing room at the start and end of the description.
Here’s a simple method to measure gaps:
- Import your video into an audio editor or video editor with a waveform display.
- Zoom in on the timeline so you can see the dialogue clearly.
- Identify the end of one line of dialogue and the start of the next. The gap is the time between them.
- Subtract a 0.2-0.3 second buffer from both ends to avoid clipping into the dialogue.
- The remaining time is your usable window.
If the gap is too short, you have three options: shorten the description, speed up the narration slightly (but not so much that it sounds unnatural), or, if you have the rights, extend the gap by a few frames in the video edit. The latter is rarely done because it alters the original pacing.
Writing Descriptions That Fit
Writing for AD is a skill. The description must be factual, objective, and concise. You describe what is seen, not what you infer. For example, instead of “She looks angry,� you say “She frowns, clenching her fists.� The description should also prioritize the most important visual information. If a character is holding a gun, that is more important than the color of their shirt.
To fit the gaps, you must write to the time. A common technique is to write the description first, then time it by reading aloud. If it takes longer than the gap, you trim. Here are some guidelines:
- Use active verbs and avoid unnecessary adjectives.
- Do not describe things that are already obvious from the audio.
- If a scene has rapid cuts, you may need to describe the overall action rather than every detail.
- For text on screen (like a sign or a phone message), read it verbatim if it is important, but only if it fits.
A practical example: a scene shows a character walking into a café, looking around, and spotting a friend. The dialogue gap is 3 seconds. A description that fits: “He scans the room, spots Maria, and waves.� That’s about 8 words, which fits in 3 seconds if spoken at a moderate pace.
Recording and Mixing the Narration
Once the script is ready, you record the narration. The narrator should speak clearly and at a consistent pace. You want a neutral tone, not overly dramatic, so it doesn’t compete with the actors’ performances. The recording should be done in a quiet environment with a good microphone. For a simple setup, a USB microphone and a pop filter are sufficient.
When mixing, you place each narration clip at the start of its designated gap. The volume should be set so the narration is audible but doesn’t overpower the original audio. A common practice is to duck the background music or sound effects slightly under the narration, but never duck the dialogue. You can use sidechain compression in your DAW to automatically lower the music when the narration plays, but that’s an advanced technique.
For accessibility, the final video should have the AD track as a separate audio track (e.g., a secondary audio program, or SAP) so viewers can turn it on or off. On platforms like YouTube, you can upload a separate audio track as an “audio description� track. For broadcast, you might deliver a file with the AD mixed in or as a separate element.
Common Mistakes and How to Avoid Them
Several pitfalls trip up new AD creators:
- Overlapping dialogue: The most common mistake. Always double-check the waveform to ensure the narration ends before the next line begins.
- Describing the obvious: If a character says “I’m leaving,� you don’t need to say “She walks to the door.� Only describe what is not in the audio.
- Reading too fast: If you have to rush the narration to fit the gap, it sounds unnatural. Better to trim the description.
- Ignoring sound effects: Sometimes a sound effect is important (e.g., a gunshot). Don’t talk over it. Wait for a gap that doesn’t have crucial sound.
- Not using timecodes: Without precise timecodes, the mixing stage becomes guesswork. Always mark the start and end of each description.
To avoid these, always do a full playback review with the AD track enabled. Listen with headphones to catch any overlaps.
Tools and Alternatives
You don’t need expensive software to start. For a free pipeline, you can use:
- Audacity for recording and editing narration.
- DaVinci Resolve (free version) for video editing and mixing multiple audio tracks.
- YouTube Studio to upload the AD track as a secondary audio track.
If you need more advanced features, there are professional AD tools that help with timing and script management, but they are not necessary for basic projects. Another alternative is to use text-to-speech (TTS) voices for narration, but the quality is usually not acceptable for professional work. For a quick test, TTS might work, but for a polished product, a human voice is recommended.
Frequently Asked Questions
What if there are no gaps in the dialogue?
If the dialogue is continuous with no pauses, you have a few options. You can try to shorten the description to a few words that fit in a very short pause, or you can place the description over a less important sound effect. In some cases, you might need to accept that you cannot describe everything and prioritize the most critical visual information. For example, in a fast-paced action scene, you might describe the overall situation rather than every punch.
How do I know how long a description takes to speak?
Read the description aloud at a natural pace and time it with a stopwatch. A good rule is that the average person speaks about 150 words per minute, so a 10-word description takes about 4 seconds. But always test with your narrator, as pace varies.
Can I use automatic speech recognition to generate descriptions?
Automatic speech recognition (ASR) can transcribe dialogue, but it cannot generate descriptions of visual content. There are some experimental AI tools that can describe images, but they are not yet reliable enough for professional AD. You still need a human to decide what is important and write concise descriptions.
Do I need to describe every scene?
No. You should only describe scenes that are not fully understood from the audio. If a scene is just two characters talking with no important visual action, you might not need any description. The goal is to fill gaps in understanding, not to describe every frame.
What is the difference between audio description and subtitles?
Subtitles (or captions) are text on screen that represent dialogue and sometimes sound effects. Audio description is narration that describes visual elements. They serve different purposes: subtitles are for deaf or hard-of-hearing viewers, while AD is for blind or low-vision viewers. Some content includes both, but they are separate tracks.
How do I deliver the final AD track?
For online platforms like YouTube, you can upload an additional audio track and mark it as “audio description.� For broadcast, you typically deliver a separate audio file (e.g., an AAC or WAV file) that the broadcaster can mux into the stream. Always check the requirements of your distributor.
Related
Method: the standard formula for this calculation, rounded as each result states. See how the calculators work or report an error.