Why did we start developping Transcriber?

In the framework of a project for automatic transcription of Broadcast News in French, we needed a corpus of manually transcribed Broadcast News material to train and develop the system. For that purpose, a suitable software environment was required. The LDC already had the experience of creating Broadcast News corpora. However, the software environment they use ran on high-end servers or rely on third-party software. We therefore looked for other solutions (which are detailed in an article presented at the LREC conference; see Question 3 below), but none was fully satisfactory, and we therefore decided to build our own tool, in collaboration with the LDC.

Why did we release Transcriber as Free Software?

There are several reasons:

• To encourage people to produce transcriptions.
• To facilitate sharing of corpora: Corpora developped at different sites are more likely to be consistent if they are produced with a common tool.
• Because making Transcriber widely available in source form can increase its quality: We hope that the feedback we will get in terms of suggestions, contributions and bug reports will outweight the time spent to distribute it and organize the discussions about it.
• And last but not least, because the free software model is getting more and more successful.

To understand better why one might be willing to distribute software freely, you can read about the philosophy of free software (mirror in Europe) promoted by the FSF.