Unit of cost: voice-over character, not minute of film
With synthetic voice the cost of recording grows with the characters of the text, not with the minutes of material. That is why the budget for the induction film is counted in voice-over characters rather than studio hours. The other costs, namely the script, footage from the plant and editing, depend on the plant and have no shared rate.
InnerWeb does not deliver ready-made training films: the library holds the client's material, and the induction film is produced according to the plant's script. The figures below come from producing the induction film at a corrugated cardboard packaging manufacturer, where five versions were made for five groups: employees, guests, truck drivers, couriers and external companies.
Three ways to produce the same five versions
The naive approach is to record each version separately from start to finish. Usage is then 15,712 voice-over characters for the set of five versions, which is on average about 3,142 characters per version.
The second approach is to split out segments. The versions are built from the same blocks, for example the rules for moving around the plant, and only some blocks are specific to a group. After splitting out 32 unique segments, usage drops to 10,737 characters. The saving is 4,975 characters, or about 31.7 percent.
The third approach is to split out unique sentences. When the same message appears in several segments, it is recorded once. After splitting out 193 unique sentences, usage drops to 8,370 characters. The saving against the naive approach is 7,342 characters, or about 46.7 percent.
The difference between the second and the third approach is 2,367 characters, which is an additional 15.1 percent against the starting point. It is worth recovering this when there are many versions or when the film will be updated often.
Voice-over characters for five versions of the induction film
Why splitting out sentences also pays off with updates
Saving on characters is half the benefit. The other half is the cost of change. When the plant changes the assembly point, in a film built from blocks one block is replaced in all versions, instead of recording five films again. At the plant in question this is exactly how the change was introduced, with a frame marked with the ISO 7010 E007 symbol.
Film files are added to the library as a data change on the installation, meaning a catalogue file and files in the object store, not a change to the system code. Before every edit of the catalogue the system saves a copy of the previous file, so a wrong title or category can be reverted. The administrator changes the description, thumbnail, title, number, category and order in the browser, without a developer.
Cost conclusion: the more the film is built from repeated blocks and sentences, the cheaper each further version and each correction becomes.
How to calculate the voice-over budget for your plant
The procedure has four steps. First, break the versions down into blocks and sentences. Second, count the characters of all versions separately, which is the naive variant. Third, count the characters of the unique segments and then of the unique sentences. Fourth, compare the three values and decide how deeply it pays to split.
For a plant with two versions and no update plans, splitting into segments is usually enough. For a plant with five groups, several languages and a plan for changes to the assembly point, it is worth going down to the level of sentences. When planning, remember the separate cost of working time: the script, the test and the review do not share one common conversion rate.
Before ordering production, check the balance of characters on the speech synthesis account, because language versions increase usage. The multiplier for the English version, about 1.04, is described in a separate page on the cost of language versions.
Character saving against recording separately
- Segments
- Sentences
Voice-over character usage for five film versions
| Parameter | Value |
|---|---|
| Number of film versions | 5 versions |
| Characters, each version separately | 15,712 characters |
| Characters after 32 unique segments | 10,737 characters |
| Characters after 193 unique sentences | 8,370 characters |
Result: Saving of 4,975 characters (31.7 percent) with segments and 7,342 characters (46.7 percent) with sentences
Note: Sample calculation based on the numbers from one production; calculate your own plant after breaking down your own blocks and sentences.
Video
Frequently asked questions
Why do we count in characters and not in minutes?
Synthetic voice-over is billed by the characters of the text. The length of the film in minutes depends on pace and shots, while characters determine the usage of the synthesis limit.
Does splitting into sentences always pay off?
It pays off with many versions and frequent updates. With two versions and no change plan, splitting into segments is usually enough.
Do these numbers apply to every plant?
They are numbers from one production of five versions. The result depends on how many blocks are shared between groups, so your own plant has to be calculated after breaking down its script.
Does InnerWeb deliver ready-made films?
No. The video library holds the client's material, and the induction film is produced according to the plant's script.
Let's talk about the induction film at your plant
We will show how the video library, the registration kiosk and the evacuation list connect into one entry flow. Book a conversation or a demo.
Let's talk