Fixed rejected sample - feedback requested
Hello,
I submitted a sample which was rejected for this reason:
There is a high level of white noise/hiss impacting your sample that is adversely affecting the quality.
I have applied a noise gate and cleaned up some little glitches, would you please let me know if you think this one will pass? I do not wish to be deactivated before I even get one job. :D
https://humanint.com/audio/phone-system2.wav
All the best,
Julie
Please sign in to leave a comment.
Hi Julie,
Thank you very much for reaching out, we hope you're having a great day.
We just reviewed your sample and asked for feedback, there is a slight hint of hiss noise in some parts at the ends of your phrases, but seems like we could approve it.
Out of curiosity, is this for an application project?
Ahoy there, Johnatan!
Yes, it is a good day, hope you are having a good one too.
Thanks so much for reviewing it! It helps a ton to know what I might be doing wrong.
This was for a sample to add to my profile, not an application project. I'm not sure what an application project is, but if you mean my initial Bunny application, that was already approved. I'm following your suggestion and increasing the number of my samples. Until I get a rhythm going about the Bunny approval process, I don't feel good about submitting samples blindly.
I don't hear the hiss you mentioned. Is there a way to visually see it some way, or detect it with a tool? It would be so nice to know when it is present, then I could know when I have fixed it. I would have the strength that comes from knowing.
All the best,
Julie
It was a great day and seems like today is going to be too, hope the same happens for you, Julie.
We just did something we hope can help you understand and point out that hiss noise. We took the last few seconds of your recording, where you'll be able to hear the hiss noise not only while speaking, but in the silence after your recording, then we copied that exact same part, and we added a Noise Suppressor plugin, which makes a clear contrast between having hiss and removing it from your audio.
Please give it a listen and let us know if you can spot it now. https://we.tl/t-QaZ25Fy4vo
Also, you can also experiment with some noise suppressor plugins on your DAW and see if you can replicate this result,
Let us know what you think
Hi Johnatan,
Thank you for the link but it doesn't work for me.
I trust that you found the hiss before you suppressed it. My real question is how can I find such an error, so that I will know I need to suppress it? Just to reiterate, I don't hear it.
If I knew the VoiceBunny noise parameters, I could work to accommodate them. Is -60dB hiss too loud? Is -70dB hiss too loud? I assume the head BIC (bunny in charge) has drafted a set of rules for people like you to follow. If I could get a peek at those rules, it would make your job so much easier.
Cheers,
Julie
Hey, Julie!
Thank you very much for reaching out and for all the info relayed during this convo with the amazing Johnny. We appreciate it very much. I think we can solve the mystery of this white noise once and for all, but first, a little preamble:
I noticed while analyzing the file, that this recording has one true peak of -3.50 dB, which paired with the sound signature of your recording, leads me to believe that the recording's amplitude was achieved through tracking gain (the gain level used in our preamp/interface when recording). This will cause the slightly saturated sound of your waveform, along with a perceivable spike in the self-noise of your recording setup, audible as the slight hiss present throughout this sample.
Here's the analysis of the levels of the sample:
Here's the spectral analysis of your recording helping us visualize the audible hiss:
And finally, the same regions after two passes of a noise gate at a threshold of 38 dB (with the fastest attack/release settings):
The issue with relying solely on processing (or careful editing) to remove this noise, is that while it certainly works, some evidence of it will always be present in the full dynamic range of your recording, as the excess gain used while recording will imprint that sound in your voice too. This is why I'd recommend you dial back the gain level in your preamp/interface by 40 to 50% so that your recordings print at a level of -9 to -6 dB, which you can later normalize to -3 dB and treat with a single pass of the noise gate to solve any low-amplitude noise issues, including hiss, mouth clicks, etc. This will also help with reducing the colored/saturated sound of the present sample, leaving you with a cleaner sound overall.
Here's what your sample sounds like after being treated with the noise gate and some EQ to reduce the saturation in a couple of frequency ranges, namely the 2400 and 12000 Hz bands, plus a 50 Hz low-cut and a slight boost in the 100 Hz range. For reference, here's a screengrab of my preferred noise gate settings for VO work, as used in Adobe Audition's stock plugin.
I'd be thrilled to read that these pointers worked for you, so please come back to the thread with some good news and fresh audio! =)
Cheers,
- Héctor Adolfo Ituarte (Bunny Studio QC Agent)
Hi Héctor,
Thank you for the awesome feedback and visual aids. I see the hiss, even if I don't hear it, so that is helpful. The numerical measurements you imparted are really meaningful.
What you suggested has taken me down another rabbit hole: learning about spectral analysis. It looks like the visible hiss at my 1.3-1.7-second mark in the 2k frequency is at -80dB or quieter. Pretty quiet hiss, wouldn't you say? Is there such a thing as "too quiet" for VoiceBunny submissions (I know ACX, for example, will reject files that have a noise floor that is too quiet).
The file of mine you were editing was already massaged twice before you got hold of it. So now I am evaluating how to incorporate a hiss-removal type of noise gate and the EQ settings you advised into the effects I will use on a raw recording to get the same result as all three phases of my and your tweaks.
While I pore over all that goodness, may I again encourage you Voice Bunny folks to consider releasing your "master list" of things you all look for to disqualify a piece of audio -- and these must be numerically measureable. I think it would be supremely helpful for everyone within the bunny warren to be operating with the same set of rules. (I was going to say "hutch" instead of "warren," but I think we can all agree we are not captive bunnies, but free-range individuals.)
I will post my results here, and I look forward to your next reply. Many thanks!
Julie
Hey, Julie!
Thank you for the great response, I'm glad that all of my ramblings were useful to you!
As for the level of hiss in Bunny Studio, we actually encourage our pros (especially in the most sought out categories/markets) to aim for a -90 dB or lower (which is technically the infinity range in a DAW) as most of our materials are processed down the line and the clients demand cleaner, crisper audio every day. This is not hard to achieve with the techniques I shared before and once you tune your ears to that level of true silence, you'll become dissatisfied with anything else, believe me!
As for that master list, the general guidelines indicate the aforementioned -90 dB or lower noise floor, which is basically inaudible hiss for us humans. No mouth clicks, breaths (except for those needed in an emotional read for instance), no background noises (you wouldn't believe how many people submit recordings with clothes rustling or such type of noises), the least amount of echo possible, which we realize is one of the hardest things to achieve in a home studio, but totally doable, and what I like to call 'indetectable processing': laying off compression/limiting or steep EQ curves that make the voice sound unnatural, as the audio will most likely be processed to an engineer's/mixer's taste or need down the production pipeline, and having overly processed VOs will make their life hell or may cause an outright rejection from us or them as clients. Also, try to avoid NR/NS plugins like the plague if possible, as they can be more harmful than useful in most cases.
What I recommend all pros with audio woes is to get themselves a pair of Sony MDR-7506 studio headphones. It's our standard-issue headset for QA and production because it's also an industry-standard when it comes to audio. If you can hear any imperfection, hiss, noise, sibilance, thumps, plosives, or echo on those when working on your recordings, so will we and most likely, the clients with production needs in their schedules.
As a VO pro, QA Agent and producer I tend to be very tough but also fair in my assessments, and on a totally personal note, I can tell you my biggest pet peeve: lack of attention. Many pros submit subpar audio, not due to the lack of tools, plugins or gear, but because they have so much work or something, that they start overlooking the important small details, like cleaning pesky salivation noises, slight degrees of hiss or other things that to me, indicate that they didn't put enough TLC into their final delivery.
I'm looking forward to your results! Rest assured that these wild bunch of bunnies will help you in every way possible! =)
Cheers!