Audio Digitization is a completely different type of mole hill than Image Digitization. There are different concerns, such as playback speeds, distortion, clipping, etc. There are also different types of technical metadata needed, such as track begin and end times. If done by hand, the digitization process for audio can be slow and tedious. Thankfully, the process is greatly benefited by the use of custom scripts.
Take the Rohlig collection for instance (a collection of organ music performed by Harald Rohlig). The original reel tapes are digitized into master wav files. These master wav files are then digitally optimized (volume leveled, extraneous silence removed, etc.). Then the master wav files are split into sub-item tracks. Finally these sub-item tracks are converted from wav to mp3 format. After initially preforming this process manually for several items, I decided to create a script to do the bulk of this work, thus speeding up the process and removing the possibility of human error.
I started by researching technical metadata standards for audio. I came across something called a cue sheet. It’s the type of metadata that might be hidden on a music cd so your media player could use it to determine where tracks begin and end. This type of file appealed to me because it was easy to read, easy to parse, and easy to create since it is basically a text file with a .cue file extension.
After studying up on cue sheet structure, I created a script (CueMaker) that would take a text file I supplied with track begin and end times, and use it to generate a cue file for each item. Once this was working, I created another script (CueSplitter) that would pick-up the generated cue file and split the master wav file into sub-item tracks. Once split, the script converts the sub-item tracks from wav to mp3 format and renames them correctly. Then I created a graphic user interface (Cue GUI) that could launch both scripts with the push of a button. Later, I added a script that can level the audio volume as well.
This is still an experimental process, but so far the results have been great. Last month we were able to upload 18 new items to the Rohlig collection, consisting of 207 audio tracks. There are more items still in the pipeline for Rohlig; we hope to have those online in the near future as well. You can listen to the Rohlig collection on Acumen HERE.
-Austin Dixon, Digitization Technologist | Hoole Library
2 Responses to Audio Digitization: a look at the Rohlig process