It's possible to use the WordNet workstation/server system to restore audio calls from Wordnet tapes, but the process is extremely cumbersome and time consuming. We don't use Wordnet itself to extract call log information. By examining tapes and meta-data, we can reverse-engineer the format and write our own suite of extraction tools.

We generally expect WordNet data to be organised on the tape in a repeating format with the single meta-data block 332 bytes long.

We can extract all the calls and meta-data (each tape typically logs hundreds of thousands of calls) in a couple of hours using our own tape drives, something which would take more than a week if the original WordNet software is used.

Each Wordnet system is typically customised by the installer depending on the client’s requirements. So there are often differences between installations. In our experience, the recording devices log each call onto tape in near real-time, and only cache a very small amount of data on the loggers themselves.

We can extract the calls and meta data using our own handlers. The calls can be converted into standard pcm .WAV files or highly compressed gsm .WAVs (or any other format the client requires) in any directory structure required. The meta-data is typically provided in an .xls spreadsheet with a hyperlink to each individual audio file. I would expect your WordNet data to be organised on the tape in a repeating format. Is this the case? Is the single meta-data block 332 bytes long? Are you allowed to send us a sample of the meta-data, at least?