software changes to uweAudioSegmentation
Just a quick post to say I'm working on changes to the uweAudioSegmentation software component - making steps towards modular components that plug in to a core web-based administration tool (perhaps called "uweTranscriptionHub").
At the moment, the gitHub repository for uweAudioSegmentation features a mash-up of:
- modified open-source audio segmentation software
- bash scripts
- ...and a web-interface for administration tasks
I am embarrassed to admit that the mash-up is the result of extending legacy designs to meet current needs. The inclusion of a web-based interface is part of the workflow of an earlier research project (specifically “Deep learning via student engagement with semi-automated lecture transcripts”), and although this approach has been extended to support our current workflow: it is not perfectly suited. It certainly isn't fit for general-purpose use by other educators!
The uweAudioSegmentation currently supports our workflow by:
- monitoring a shared "dropbox" for uploaded files
- detection and extraction of audio from uploaded files
- conversion between common audio file formats
- web-based administration of uploaded/extracted/converted audio
- web-based configuration of audio segmentation parameters
- audio segmentation and timecoding to TRS XML
- monitoring of transcribed text segments
- editing of transcribed text segments
- production of SRT subtitles from text segments
Many of these tasks are outrageously outside of the remit of "audio segmentation". I am beginning work to make the software more malleable, and therefore more likely to be useful to other people. I plan to accomplish this by firmly separating the tasks into individual software components, whilst collecting all of the "administration" tasks into another generalised web-based hub.
Watch this space (and my gitHub profile) for updates to the software components!