Diligence & quality
Recorded hours vs usable dataset hours: how to count media volume
File-hours, unique activity-hours, speech-hours and accepted hours answer different questions. Define the unit before comparing collections or recording offers.
- Topic
- Diligence & quality
- For
- Buyers and data owners
- Reading time
- 5 min
- Updated
File-hours, unique activity-hours, speech-hours and accepted hours answer different questions. Define the unit before comparing collections or recording offers.
Use the linked free tool to work through the practical steps. Keep the original files and documentation unchanged, and save your report with the date and scope of the review.
File-hours count the duration of files
Add the duration of every inspected recording to get file-hours. Four cameras filming the same one-hour activity can produce four file-hours. This is a useful storage and delivery measure, but it does not create four hours of distinct activity. Alternate encodes and duplicate exports can inflate the total further.
The Media Dataset Inspector labels its sum as known file-hours. It excludes unknown durations and shows how many files contributed a duration. Do not describe an incomplete sum as the complete archive length.
Unique activity needs session information
To calculate unique activity, group files by recording session, participant or event and align their time intervals. One long video may also contain breaks, setup or repeated takes. Metadata alone does not identify those intervals.
Retain a session identifier and capture-time convention in a separate manifest. For multiple cameras, document view names, synchronization method and missing coverage. Do not divide a total by the camera count unless all views cover exactly the same intervals.
Speech-hours require a speech measurement method
A one-hour podcast file can contain silence, music, overlapping speech and an intro. Recorded duration is therefore different from speech duration. If voice activity detection is used, name the system and settings, and review performance on the actual acoustic conditions.
Decide whether overlapping speakers count once as an elapsed interval or separately as speaker time. Both measures can be useful, but they must not share an ambiguous label such as total audio hours. The metadata inspector does not detect speech or assign speaker identities.
Accepted hours follow the written specification
Accepted volume depends on the task and agreed tests: coverage, permissions, image quality, synchronization, annotation or other requirements. A raw recording count does not predict acceptance. Separate recorded volume, submitted volume and accepted volume in a delivery ledger.
When comparing offers, ask which unit is billable and how exclusions are measured. Use the existing Fiund recording earnings calculator with your own written rate and acceptance assumptions. Its outputs are scenarios, not market valuations or guaranteed compensation.
Sources and further reading
Try the workflow on your own data
Open free tool →Related resources
Related reading
Open the free tool
Browser tools
Free dataset tools
Inspect local media, document dataset scope or check an evaluation sample.
Related reading
Dataset quality and delivery
5 min read
How to inventory an audio or video dataset
Build a file-level inventory before estimating volume, requesting a quote, or preparing a dataset for review. Count known durations and record the technical properties buyers need.
5 min read
How to write a dataset card buyers can evaluate
Document the dataset’s purpose, origin, collection process, limitations and permissions evidence. A useful card makes unknowns visible instead of hiding them behind marketing language.
5 min read
Dataset manifest vs datasheet vs dataset card
A manifest identifies files. A datasheet records structured answers about a collection. A dataset card explains the collection to users. Most serious deliveries benefit from all three.