Disagree with revision. (Volume and breath problems.)
I sent the email below to the Voicebunny support team and they asked me to post my question here. So here it is:
I have been asked to do a revision for the Amp – Marvel 2.0 – Dutch project. (Project ID: 359911B915E27EFC82AB)
These are the instructions:
Thank you for this submission! Unfortunately, there are some issues to address:
1) Breath Control: Parts of the script with more than one sentence present audible breaths at the start of each sentence. This is a distracting and undesirable quality that can be corrected through careful editing.
2) Levels: Your audio files are not normalized to the -3dBFS peak level Bunny Standard. Please make sure to normalize your files before resubmitting.
About question number one: I left the breathing in on purpose as this is a (sort of) long form audio project (some short one line parts, but also 20 lines+ for an e-learning module) and I think taking away the breathing in between makes the takes unnatural. Also: I already submitted a similar project, for the same client, that will go to the same target group/audience, right before this one, and I left the breathing in. This project was approved. A couple of months ago I did another 4 projects for this employer, same purpose, also with the breathing (in between lines) left in. So I think for consistency it’s also important to leave them in. The client seems to agree, because they approved the previous 4 projects I did for them as well.
About question number two: as I’m doing multiple projects for the same employer, I’m trying to keep the levels of the audio consistent (and as close to the -3dB threshold) throughout all projects and all separate files. If I’d just normalize all my separate files to -3dB, the perceived audio loudness would be very different from file to file, as every file has slightly different peaks. Maybe you already know this but I’ll explain it to be sure:
An example case: File A has a (pre-normalizing) peak of -12dB and file B has a (pre-normalizing) peak of -6dB because of one syllable I said a bit louder or because the fragment has a more enthusiastic piece in it. The perceived loudness of both are the same, and relatively correct, as I recorded them with the same pre-gain, studio, mic, mic distance etc.
If I normalize both files to -3dB, file A will get 9dB of gain and file B will get 3dB of gain. In that case there’d be a 6dB difference in perceived loudness. This provides the listener with a very inconsistent listening experience, and he/she will have to adjust the volume constantly throughout using the manual/website/guide etc.
The way I solved it: I looked at the highest peak in all the files in all projects. In the example -6dB in file B. And then I used 3dB of gain (to get to a peak of -3dB in file B, the file with the highest peak) on all audio files. This does mean that file A ends up with a peak of -9dB, but it also means the listener gets a more consistent and comfortable listening experience overall. I think this is the best solution for these projects.
So the question: do you agree? Shall I submit the revision without making any actual changes? Or do you still want me to make changes? If so: which ones?
Kind regards,
Bob
P.S. Because the deadline to send the revision is coming up I've submitted the revision without making any changes. If you do require changes please send another revision request.
Please sign in to leave a comment.
Hi Bob,
Thanks for sharing this case with us. The project has now been approved.
As you say, normalization is not a perfect solution in all cases. As we receive thousands of audio files all with different volumes, we try to aim for -3dB in order to provide some sort of consistency for our clients and an easy process for Pros to apply. As you well pointed out, it's better to provide consistent loudness but it would be harder to ask that from Pros (and to review it). I'll forward the case to the team so they can learn from it.
Hi Bob!
I'm the ogre who asked you for that particular revision. I must say you're knowledgeable and do concur with my colleague Sebastian on his opinion, while also reinforcing the concept of normalization vs. loudness:
I know that it's a better, more elegant solution, to achieve a perceived loudness and to use the LUFS standard, but believe me: not all pros are as audio/tech savvy as you are, thus we need to settle for a good ol' -3dBFS normalization standard for the time being.
As for the breaths, I understand they might seem as a "stylistic" choice where some breaths are left in a read by design where they would be expected. In your case, however, I noticed loooong breaths corresponding to that first grasping-for-air we do before recording a sentence being left at the start of most files in that particular submission. I had the chance to work on another part of this project and that situation did get better!
Now, a lot of clients do not mind and in fact do not use our files as-is: editing, processing and splicing the audio to whatever their desire is. This just might be the case. Other times, clients can complain about us not delivering a "ready to use" audio file, and rightly so: it's a part of our promise in VoiceBunny.
I'm not a director and wouldn't tell another talent how to do their job or to modify their style, so take these words with a grain of salt: it's just a friendly pointer from one pro to another on what will work most of the time, as opposed to something that may lead to a client rejecting a project even if we approve it internally.
So: keep up the good work! You're a professional and committed VO artist and we're super glad to have you on board!
Cheers,
Héctor Adolfo Ituarte (VoiceBunny QC Team)
Hi Héctor,
Haha well, for an ogre you seem very friendly and knowledgeable yourself.
Thank you for taking the time to explain in such detail, I understand your arguments and I will make sure to learn from them.
Have a good day!