Tuesday, January 2, 2018

Toujours de l'audace (Audio Editing)

Over the years, I've discussed many of the skills needed for fansubbing: translation, timing, editing, typesetting, QC, and encoding. In 2017, I encountered a new one: audio editing. It's not often needed, but when it is, the situation is usually fairly dire.

I ran into two examples last year. The first case was in the fourth and final episode of Akai Hayate. For some reason, the ending song - and only the ending song - was horribly distorted on the laserdisc release. In comparison, the VHS version sounded much better, but it was only in mono. It was impossible to use the endings from the first three volumes, because the fourth volume's ending was unique. The second case was in the complete release of A-Girl. Due to an incorrect setting, tape stretch in the video resulted in bad audio, with the sound audibly dipping in tone. A second, audio-only rip, done with the right settings, sounded better but had different volume levels and slightly different playing times.

In both cases, the solution required splicing together different audio sources and then adjusting for differences in volume and tone. Enter Audacity, a full-featured, open source audio editor. Audacity can do almost anything, but its user interface is not particularly intuitive. Fortunately, the Internet has many, many write-ups on how to use Audacity for various tasks. Simply Googling "Audacity <do this task>" will find the right answer fairly quickly.

For both Akai Hayate 4 and A-Girl, the basic procedure was the following:
  1. Load both audio tracks into Audacity.
  2. Use the Time Slip tool to align the new audio track with the old one.
  3. Splice an appropriate section from the new audio track over the old one.
  4. Use the Amplify effect to adjust the volume of the new section to match the old one.
  5. Export the spliced audio as a WAV file.
  6. Use an external encoder to transform the WAV file to the audio format of choice.
Now, there are a couple of trivial and non-trivial annoyances in getting this done.
  1. The Time Slip tool doesn't have a fine adjust. All aligning is done with the mouse, and the mouse doesn't have enough resolution to get to a sample boundary, even at really high magnification. It would help if the tool responded to the arrow keys for fine-grain motion.
  2. Selecting a section of audio is clumsy. You can use the mouse to define the start - and even fine-adjust the start with the arrow keys - but the end is defined by dragging the start across the selection. Aegisub's system of allowing independent definition of the start and end points works better.
  3. There's no straightforward "spice and replace" function. Instead, there's a complicated series of arrow keys to get back and forth between old and new. Aegisub has a "paste over" function for replacing sections of a script, and that works better.
Still, there's a lot that Audacity will do automagically to make things work. For example, when it was time to export the spliced audio track for Akai Hayate 4, Audacity noticed that the spliced area was mono and duplicated the track to create a 2-channel (but still mono) mix.

So if you're a fansubber looking to expand your bag of tricks, I recommend spending some time fooling around with Audacity. You never know when it will come in handy.

No comments:

Post a Comment