QC on Project 32079367C49AE06A8851
Hi, everyone.
Been hearing some strange stories about Quality Control lately and wanted to share my experience.
Two nights ago I had a Contest Backup entry rejected because:
"Unfortunately, your audio does not sound like the sample chosen by the client. The reference is a more vibrant, energetic and happy sounding voice."
I took a second look at the "Additional instructions" in the Client's prompt and don't seen anything about vibrancy, energy, or happiness. Additionally, here is the Sample Chosen By the Client:
https://www.dropbox.com/s/buhrtwus9p5mxvk/Sample%20Preferred%20By%20Client.mp3?dl=0
This is a fine recording, but I wouldn't characterize it as vibrant, energetic, and happy. It's more conversational than anything, very matter-of-fact. More importantly, though, do you really want your Bunny Managers subjectively interpreting the vocal qualities of referenced samples? And, if so, are you sure this Bunny Manager's judgement is on point?
Beyond that, I want to play you a contest entry that was accepted under these standards:
https://www.dropbox.com/s/qxeyi2n3bcueb06/Approved%20Entry.mp3?dl=0
While this entry may have its own merits, do you think it sounds like the sample preferred by the client? Do you think it is "vibrant, energetic, and happy"?
Finally, here's my rejected entry:
https://www.dropbox.com/s/qg73vwzbxt3c575/Rejected%20Entry.wav?dl=0
I'm not posting all this to make you guys look bad. I love the platform and am having fun with it. But you may want to reconsider the way you're training VoiceBunny's QC agents. When a client provides few additional instructions, are you certain that Bunny Managers should be inferring the vocal delivery desired by a client? And are you comfortable with the way they've done so in this case?
Because it's giving many of us pause.
Thanks for listening!
Scott
Please sign in to leave a comment.
Hi Scott,
Thanks for sharing this case with us. I agree the language is not entirely accurate in this case, but overall the sense of the rejection is. The main point is that your submission doesn't sound close to the sample. The sample sounds higher pitched, younger (your tone is more gravelly) and overall your delivery is almost whispered at times. This is not to say it's a bad performance, just that it doesn't entirely fit with what the client is looking for.
I'll share this case with the team, as I believe a revision would have been a better course of action. Also, we'll work on improving the language we use to explain these types of cases.
Thanks again fro sharing,
-Seb
Thanks for the reply, Sebastian.
The larger concern though is whether these contest entries were judged against the same tonal standards. Have you considered using a single QC agent for each contest? Might help.
Hahaha, wow, Quality Control is worse than I thought. Your voice over is infinitely better than the previous two. Not even close. They have mouth noise, weird noises here and there, and it sounds like they were recorded on rather mediocre microphones. The only problem with yours is that it may be slightly over compressed or maybe normalized at a level higher than should be, but those are easy fixes. Could use a de-esser as well.
I'm with you, QC leave a lot to be desired. Then again, I've been an audio engineer for 10 years.