Choosing a Voice for Audio Description: An Accessibility Checklist

Start with the purpose of audio description

Audio description communicates relevant visual information through spoken language, helping blind and low-vision audiences follow a film, television program, performance or other visual work. Its purpose is not to narrate every visible detail. It supplies information that would otherwise be unavailable through the soundtrack, including actions, expressions, changes in location and important visual relationships.

Voice selection should therefore begin with the program's descriptive requirements. A fast-paced action sequence creates different demands from a quiet dramatic exchange. A documentary may require precise delivery of visual terminology, while a family animation may depend on clear character identification and careful timing around dialogue. The intended audience and distribution environment also matter.

• Identify the program's genre and intended audience.
• Determine how frequently descriptions must fit between dialogue.
• List names, specialist terms and recurring character references.
• Consider television, streaming, theater or mobile listening conditions.

Intelligibility, pronunciation and vocal consistency

An effective description voice makes information easy to understand on the first listen. Clear consonants, stable vowel production and controlled articulation support intelligibility, particularly when a description contains unfamiliar names or technical vocabulary. However, excessive enunciation can sound unnatural and consume valuable time. The objective is clear speech at a pace appropriate to the material.

Pronunciation consistency becomes especially important in programs with recurring locations, characters or specialized terminology. Create a pronunciation reference before production and identify words that may require phonetic notation or a specific pronunciation convention. When evaluating a synthetic voice, check whether its pronunciation controls support the required vocabulary and whether the same name remains consistent across separate passages.

• Check proper names and recurring terminology.
• Listen for consonant clarity at the intended playback level.
• Assess vowel quality and word stress.
• Check pronunciation consistency across separate recordings.
• Avoid a delivery style that sounds artificially emphatic.

Pacing and the available descriptive window

Audio description often occupies spaces between dialogue and important sound events. The voice must deliver enough information within those windows without rushing. A description that is technically complete but difficult to follow at normal playback speed does not serve its purpose. Teams should examine both the duration of each description and the amount of information it contains.

For example, consider a 7-second window between two lines of dialogue. A concise description of a character entering a room may fit comfortably, while a sentence containing several actions, clothing details and location changes may require editing. Measure the spoken duration and replay the scene with its original soundtrack. The description should preserve the priority information without masking an important line or meaningful sound.

Voice character and the surrounding soundscape

The preferred vocal character depends on the production. A restrained voice may suit a documentary or dramatic film, while a warmer delivery may fit material intended for younger audiences. Pitch alone does not determine suitability. Articulation, vocal weight, intonation, speaking rate and perceived distance all influence how the description sits within the soundtrack.

Teams should evaluate the voice in context, not only through isolated headphones demonstrations. Background music, dialogue, environmental effects and playback equipment can alter perceived clarity. A voice that sounds balanced in a quiet recording booth may become difficult to understand over dense music. Listen at the intended delivery level and consider how the voice interacts with the mix across contrasting scenes.

• Check whether the voice competes with dialogue.
• Listen for intelligibility over music and environmental sound.
• Assess emotional restraint in sensitive scenes.
• Consider whether vocal tone remains suitable across the full program.

A practical selection checklist for accessibility teams

A structured selection process makes feedback easier to interpret. Prepare a set of representative passages covering quiet scenes, rapid action, dialogue-heavy sequences and moments with important environmental sounds. Use the same descriptive script when comparing voices. This controls the linguistic material and allows listeners to concentrate on delivery characteristics.

Evaluate each passage using separate criteria. A voice may be clear but too slow for a short descriptive window, or expressive but inconsistent when pronouncing recurring names. Record observations with timestamps and specific examples. Include listeners with relevant accessibility experience in the evaluation process, and provide a way to report comprehension problems that may not appear in production notes.

• Intelligibility at normal playback volume.
• Accurate pronunciation of names and specialist vocabulary.
• Appropriate pacing within available dialogue gaps.
• Consistent vocal character across scene types.
• Compatibility with music, effects and dialogue.
• Reliable delivery across the intended distribution format.

Synthetic voices, production workflows and final decisions

Synthetic narration can offer production flexibility when descriptions require revisions, alternate language versions or frequent updates. Its practical value depends on the available pronunciation controls, timing adjustments, voice consistency and export options. Teams should confirm supported audio formats, licensing conditions, language coverage and the ability to revise individual passages without disrupting the rest of the program. These details vary by provider and may change over time.

The same accessibility criteria apply whether the voice is human or synthetic. A convincing demonstration does not establish that a voice will work throughout a full production. Consider a representative sequence that includes at least three different scene conditions, then evaluate the completed mix rather than the narration file alone. Document the selected delivery characteristics so future episodes and revisions can maintain a consistent listening experience.

Questions readers ask