Easy to read.
Even better to hear.

Long, nested sentences may be clear and efficient for readers. Readers can go back and read a sentence again. Listeners need to understand it immediately. Divide information into manageable units and set priorities where needed.

Reading text

The new platform, which was initially tested in selected markets last year and officially introduced in spring after several technical adjustments, is intended to help companies organise their internal processes more efficiently.

The sentence is grammatically correct. When listening, however, the audience has to retain several details before reaching the main point.

Spoken text

The new platform is intended to help companies organise their internal processes better. Initial tests took place last year. Following several technical adjustments, the platform was officially introduced in spring.

The information is retained, but presented in an order that is easier to follow. A useful guideline is one idea per sentence. Additional details should follow rather than interrupt or obscure the main point.

Listening texts should also be easy to pronounce. A sentence full of repeated consonants and dense specialist phrasing may be technically precise but difficult to deliver fluently. Keep both the listener and the speaker in mind.

Texts that work well for speaking and listening usually consist of manageable units of meaning. But short sentences alone can also be overdone, as this example shows:

The project starts in September. The team is ready. The equipment has been ordered. Training is underway. There are still open questions. The schedule remains unchanged.

The sentences are short, but their relationships are unclear. What is the team checking? Which information is missing? Why does approval come later? Can the deadline still be met?

The project is expected to start in September as planned. The team is already preparing, the equipment has been ordered and initial training is underway. Some questions remain open, but they are not currently expected to affect the schedule.

Word count alone is not decisive. It matters more whether a sentence has a clear direction and is placed appropriately. Use clear references, a logical order and recognisable connections.

Numbers become difficult in listening texts when several follow in quick succession. Readers can consult tables, compare figures and calculate. Listeners find it harder to retain many percentages, dates and amounts. Simplify the information where appropriate.

Reading text

Revenue rose from €12.73 million in 2023 to €15.4 million in 2024, an increase of 20.97 per cent.

Spoken text

Revenue rose significantly within a year: from around €12.7 million to €15.4 million. That represents an increase of 21 per cent.

Precision may be essential, for example in technical instructions. Elsewhere, not every decimal place matters. Where understanding is more important than exact detail, figures can be rounded to values that are easier to grasp.

Reading text

Approximately 4,952 preregistered spectators attended the launch of the new cruise ship with its 179,248 gross register tons.

Spoken text

Almost five thousand preregistered spectators attended the launch of the new cruise ship, which measured nearly one hundred and eighty thousand gross register tons.

The shipping example uses an outdated measure of vessel size. Specialist texts may contain abbreviations such as BRT, GT or BRZ, which refer to different terminology and should not be treated as interchangeable without checking.

Even a reading text should explain unfamiliar abbreviations, especially for a non-specialist audience. This is more important still in a listening text.

Should an abbreviation be pronounced as a word or spelled out? If it is spelled out, which language should be used for the letter names? IBM, for example, can be spoken using English or German letter names.

Familiar abbreviations may have an established pronunciation, such as NATO and EU in German. Less familiar ones need guidance. A listening text referring to data sent via an API to a CRM system should introduce the terms and specify their pronunciation.

For example:

The data is transmitted through an application programming interface, or API, to the customer relationship management system, known as CRM. [Specify the required language and whether each abbreviation is spoken as a word or as individual letters.]

Once introduced, abbreviations can be used later in the text without explaining them again.

Whether numbers appear as digits or words in the script partly depends on the production process.

Professional voice talent can often work efficiently with forms such as 24%, 20 cm and 30°C. For text-to-speech systems, writing out numbers and units can be more reliable. For 125 km/h, explicitly decide how the unit should be spoken rather than leaving it to chance.

When several decimal places must be retained, the script needs to specify how to pronounce them. In German, 27.35 could be spoken as the equivalent of “twenty-seven point thirty-five” or “twenty-seven point three five”.

Dates such as 03.04.2026 also allow different spoken forms in German: numeric day and month, or a named month, and different ways of speaking the year. Decide the intended form. For English, also resolve whether the date follows day-month or month-day order.

Numbers do not need to be written out throughout every audio script. Identifying difficult points and resolving uncertainty before production helps avoid unnecessary work and costs.

“She said he was entirely innocent” uses the same words in the same order as “She, said he, was entirely innocent.” Punctuation changes who is speaking and who is described.

The commas create a completely different meaning. Consider another example:

“The minister says the journalist spreads false information.” “The minister, says the journalist, spreads false information.”

Perhaps both are right. Correct punctuation is necessary for clarity and unambiguous meaning, particularly when a text is the basis for a voice recording.

Before texts become audio scripts, they often leave unresolved whether a name or term should be pronounced in German, English or another original language.

A recording requires a decision. Should Adobe follow German pronunciation or an English-style pronunciation? How should “cache” sound? In English, it sounds like “cash”.

Pronunciation notes resolve these problems. Correct pronunciation from the start can avoid correction rounds, delays and additional costs.

A pronunciation note must be practically useful and understood by the speaker. Specialist phonetic notation is not suitable or necessary for every production team. An easily understood phonetic spelling is often sufficient.

The International Phonetic Alphabet, or IPA, may also be used for professional and multilingual productions. If IPA is the only guidance, the voice talent must be able to interpret it.

Text-to-speech systems may support pronunciation dictionaries or phonetic markup. These should also be checked before being placed into the production script.

Personal names deserve particular care. If pronunciation is uncertain, check reliable sources. Mispronouncing a name can sound unprofessional and may be perceived as disrespectful.

Consistency matters: names should sound the same throughout, even when different voices are used.

Video voice-over productions also require professionally prepared audio scripts.

For a finished video edit, scripts generally need timecodes. They are especially important when recordings are produced in several languages. Available time windows may create new requirements for the translated text.

A fast-edited video with an effective English or German script does not mean literal translations will fit the same windows. English-to-German translations may need more time at a comparable speaking pace; other language pairs can create further constraints.

If neither recutting the video nor unnaturally fast delivery is acceptable, the translated script needs adaptation.

Problems in audio are often blamed on the voice, recording equipment or text-to-speech system. The cause may instead lie in the script. Here is another example:

Reading text

During the test phase launched on 15 September 2025, a total of 1,487 records from five different source systems were migrated, with 94.6 per cent of the content transferred without manual reworking. The remaining 5.4 per cent mainly concerned incomplete addresses and inconsistently formatted product names.

The paragraph is informative, but several numbers arrive in succession and the main finding appears late. Its phrasing and density make it difficult for listeners to retain everything.

Spoken text

The test phase began in mid-September twenty twenty-five.

Nearly fifteen hundred records were transferred from five different systems. The main result: around 95 per cent of the content could be transferred automatically.

Only a small proportion needed manual reworking. This mainly involved incomplete addresses and inconsistently written product names.

An audio script does more than provide text for recording. It needs greater immediate clarity than an ordinary reading text. It should prevent misunderstandings, use punctuation correctly, simplify numbers and units appropriately, specify pronunciation and provide exact timecodes where needed.

The quality of an audio script shows in how easily the audience can follow it. Recording an unadapted reading text leaves much to chance. Editing for listening helps whether the production uses a person or a synthetic voice.

We hope these suggestions help you prepare a high-quality audio script. We can support the planning of media productions and localization from an early stage.

We support editing, translation and preparation of production-ready scripts for voice recordings, voice-over, subtitles and text-to-speech.