Crackling using Blue Yeti

Hi everyone,

I’m currently recording an audio book using a Blue Yeti USB mic, the latest version of Audacity, and a MacBook Air. On two occasions, most recently yesterday, my recording has had some artificial inconsistent crackling. Obviously noise removal can not clean this up and the audio is now useless. I want to do everything I can to avoid encountering this crackling again? What is the issue here? The crackling is not ambient since previous recordings (and the first 20 mins of yesterday’s recordings) do not pick up any crackling. Is this an issue with the position of the mic, a bend in the USB cord, something else?

Thank you for any help!

This your first AudioBook? We have some testing tools to save you a bunch of work with ACX-AudioBook compliance.

There is a “normal” failure for this and we can try those solutions, but that doesn’t appear like normal data failures. It sounds like electrical interference from a bad motor, noisy air conditioner, bad lamp dimmer, something like that. It’s electrical, not acoustic.

You use your Air portable all the time, right, taking it to lunch and to meetings and later, settle down, plug in the Yeti and record. I would start with changing out the USB cable. I think in some of those instances, the cable didn’t make good contact when you plugged it in. That turns the Yeti and cable into an antenna for any electrical interference in the house.

Yertis have no natural way to reject electrical interference that the higher end microphones do. It depends on the computer, cable and environment working perfectly.

Troubleshooting erratic problems is a real joy. The developers have a phrase “Moonphase Problems.” Sometimes, when the moon is one third waning…

Make a recording while you gently wiggle the USB connection, first one and and then the other. Announce what you’re doing. “I’m wiggling the Air end now.”

Do the noise instances correlate with the times you recorded with the Air on its charger?

Koz

That’s not a Yeti Pro, is it? Those are very different microphones.

Koz

Koz,

Thanks so much for these suggestions. I’ll go through these steps tomorrow when I can get to my recording space. The mic is indeed a Blue Yeti Pro. I apologize for not indicating that initially. Out of all your suggestions, I’m inclined to believe the noise is stemming from the USB cord as you may have suggested. I’m recording in a college practice room, sound reinforced. No dimmer. Just some humming room tone from a ventilation system. I’ve recorded in this room on 5 separate occasions with pretty a pretty consistent set up. However, I move my mic and laptop back and forth between home and the space between sessions. I’ll report back tomorrow after trying your USB suggestions. In the meantime, could you please post a link to the ACX compliance tools you mentioned? This is indeed my first audiobook.

-Logan

Probably one of the handiest things we ever did was build a tool that mimics the Evil ACX Robot which is the first thing you hit when you submit to ACX for Acceptance. Will developed ACX-Check as a handy one-pass tool out of older programs and tools. It automatically tests your presentation for the three important sound values which I printed in a posting to someone else.

https://forum.audacityteam.org/t/measure-between-23db-and-18db-rms/32770/16

This is the actual tool. Download and install it in your Audacity plugins folder. It will appear as Analyze > ACX Check.

http://wiki.audacityteam.org/wiki/Nyquist_Analyze_Plug-ins#ACX_Check

As it says in the disclaimers, it is possible to fake it out. It depends on at least a half-second of non-speech, quiet room tone to tell you what the noise is. If there is no such gap, it will guess at it and give you a wildly incorrect answer. It will cheerfully analyze a stereo track, but you will not be so cheerful trying to read it, particularly if you miss. It will create a pile of data carefully explaining what you did wrong.

Almost all voice work is mono (one blue wave) and not stereo. That’s a good deal simpler to read.

Top three values and sentence 2/3 of the way down.
Screen Shot 2016-07-08 at 5.31.49 PM.png
I wrote a piece on how to submit voice test clips to the forum.

http://www.kozco.com/tech/audacity/TestClip/Record_A_Clip.html

That tells us everything we need to analyze your work. Oddly, the hardest thing for people is the 2 seconds of silence. That is not the cue to check your FaceBook feed, re-arrange the silverware on your table or scratch your tush. You need to freeze in place and let the room perform for us.

As we go.

Koz

@Bhodi.
I’m going to ask a senior forum elf to move your posting into its own topic rather than piggyback. Stacked topics are very hard to manage

Koz

Koz,

I’ve been using that ACX check tool for a couple weeks now as I play around with mastering my chapters to meet ACX requirements. The tool is god sent and a must have for any ACX narrators! My follow up is this (and I know this is asking for a lot) : In keeping with the theme of ACX requirements, do you have any compression/EQ templates that will take raw spoken word and format the file into one that passes the ACX check? I’m finding that I have to play around with the files quite a bit to get them to meet those requirements. My biggest issues is achieving the proper RMS levels and simultaneously getting the file to have the proper peak levels. Currently using noise removal, Chris’s Compressor 1.2.6, and normalizing to -3.1 db to master my raw spoken word. I’ve scoured youtube videos to see what people’s settings are for these effect tools but there’s not a whole lot of info out there pertaining to the specifics of these settings. Perhaps this is because everyone uses different mics, records at different levels, and has different voices? If that is the case and there are not “one-size fits all” templates, I’m happy to submit a sample of one of my edited yet unmastered chapters.

Thank you so much for you help! I’m amazed at the quality of the responses and the amount of help you guys provide on this forum!

-Logan

In keeping with the theme of ACX requirements, do you have any compression/EQ templates that will take raw spoken word and format the file into one that passes the ACX check?

One button processing!!

No. Because everybody starts out in a really, really different place. I was commenting about this to someone else in a parallel message. ACX has videos explaining how to produce audiobooks and they all start with: After you install your soundproof room.

If you don’t start with a soundproof studio, your recordings are likely to be profoundly unstable, unsatisfactory and different.

Most forum work is not finishing, and polishing. It’s disaster recovery. The longest message thread in the forum is Ian in Hollywood. He lives (or did at the time) near the noisy intersection of La Brea and Venice. 39 forum chapters and months later, we got him in good enough shape to submit, successfully, for AudioBook publication. He did it by turning his broom closet into a studio. He learned to work around the air conditioning problem (Sunny Southern California, right?)

Further, you’re not correct that it always takes a list of corrections to produce a compliant clip. I have a soundproofed room and twice now using different, basic equipment I have recorded a sound test, cut off the two ends, changed the volume, checked it against ACX-Test and gone to make tea. That’s it.

Even if I miss it by not being a performer (I’m not), my corrections are likely to be miles different from someone trying to announce an audiobook with a Blue Yeti on the kitchen table.

So what you/we really want is a software package that automatically analyzes the sound, generates the proper corrections, applies them, and checks the work [dusting off hands].

Programmers/developers can apply any time. It doesn’t pay well.

We should also remember that ACX basic audio is only the first test. It still has to sound natural and pleasant to the Human Quality Control that follows. This is where the recording-in-a-bathroom submissions and cellphone voices die.

If this was easy, anybody could do it.

You are welcome at any time to submit a voice test.

http://www.kozco.com/tech/audacity/TestClip/Record_A_Clip.html

You don’t have to submit a fresh clip. You can cut down an existing performance if you have one. It should have at least 3/4 second of room tone in it somewhere or ACX-Test will fail. No fair generating silence.

We will try to make the minimum corrections to get you past the gate, and give you the list. Fair warning, if you change anything, the list is likely to be different. You can submit much longer pieces if you have a File Hosting Service outside of the forum. Remember to condition the work so we can download it. That’s required.

Koz

Koz,

I certainly don’t expect one button processing. I’ve spent hours pouring over all of the ACX blogs and video tutorials so I’m very familiar with the importance of starting off with clean room tone. The best I can do at this point, short of renting out professional recording space, is record in a sound reinforced practice room at a local university. I’m quite happy with the quality I’ve gotten. The only room noise is a constant AC hum. I’ve been able to play with my clips in the past and have gotten them to pass ACX requirements. However, based on my limited understanding of audio engineering, it’s a lot of educated guess work when it comes to compression and EQ. I have attached a short clip of one of my chapters following your instructions for submission. As I previously mentioned, I replicate these conditions every time I record, so any processing advice will be a huge help for all of my audio book endeavors.

Thank you for your help,
Logan

I’m not entirely sure where to go with that. The supplied clip has very low volume voice over a bed of high noise. It took aggressive analysis and corrections to produce a quiet, natural-sounding product.

~~

Many of the tools don’t like being forced to work in very low level sound, so the first thing I did was Normalize to -3.2. Normalize and Amplify are both just volume controls that pay attention to different characteristics of the sound.

– Select the whole clip or show by clicking just above MUTE.
– Effect > Normalize: [X]Remove DC, [X]Normalize to -3.2 > OK

I played the clip and the first several seconds of room tone are clearly too loud. I saw that by watching the Audacity bouncing light sound meter. I have the meters set to -96 (total sound range) in Audacity > Edit > Preferences > Interface > Meter Range [-96] > OK. If the tip of the bouncing light never settles to the left of -60, there’s zero chance you’re going to make noise specification.

So what’s there? Drag-select from about 1 sec to 4 seconds in the middle of the Room Tone segment. Analyze > Plot Spectrum.
Screen Shot 2016-07-16 at 5.14.14 PM.png
You may need to click the graphic or scroll to see the whole thing. Watch the bottom.

The purple stuff to the left of 20.00 is not audible by humans. That’s thunder and earthquakes. That can be sloppy manufacturing of the microphone. Everything to the left of 100 is usually not valuable and can be deleted. That’s metrobusses and trucks/lorry’s going by. Some people’s announcing voices do go down that low, so this is the first time you need to pay attention to theater.

Steve designed a handy-dandy filter that blows away everything to the left of 100.

http://www.kozco.com/tech/audacity/Documents/LF_rolloff_for_speech.xml.zip

Download that, unzip it to LF_rolloff_for_speech.xml and install it in the Equalizer tool.

Adding Audacity Equalization Curves
– Select something on the timeline.
– Effect > Equalization > Save/Manage Curves > Import
– Select LF_rolloff_for_speech.xml > OK. (it won’t open the ZIP. You have to decompress it)
– LF rolloff for speech now appears in the equalization preset curve list.

Apply it.

LF Rolloff (rumble filter)
– Select the whole clip or show by clicking just above MUTE.
– Effect > Equalization: LF Rolloff for speech, around 5000 Length > OK

Around here I selected the whole clip and Analyze > ACX Test. The noise level is still too high and the RMS (overall loudness) is too low.

About half-way home. With me so far?

Koz

Koz,

I’m following along no problem so far. I’ve normalized to -3.2db and applied the suggested downloadable Eq tool. Ready for the next steps.

-Logan

The steps until now were low-hanging fruit. Easy filters and corrections that for the most part affect ACX-Test without affecting the sound. The clip still won’t pass.

Additional noise reduction.

Noise Reduction
– Drag-select 1 sec to 4 sec.
– Effect > Noise Reduction: Profile
– Select the whole clip or show by clicking just above MUTE.
– Effect > Noise Reduction: Settings 12, 6, 6 > OK

Background noise passes now, but overall loudness still won’t. Compressor to bring the loudness up.

Audio Compressor
– Select the whole clip or show by clicking just above MUTE.
– Effect > Compressor: Thresh -20, Floor -40, Ratio 2.5:1, Attack 0.2, Release 1.0, > OK
– Effect > Normalize: [X]Remove DC, [X]Normalize to -3.2 > OK

Et voilà, the clip now passes ACX-Test.

Fair warning Noise Reduction and Compressor are running at stiffer correction than is normal and the background noise is still audible in the clip. So the clip may still fail Human Quality Control after it makes it past The Robot. Also, the values are right on the edge. If the sound quality sinks any lower, that’s the end of the world. If you throw more noise reduction in, it will give you wine glass voice. ACX hates us using any Noise Reduction for exactly this problem.

Koz
Screen Shot 2016-07-17 at 20.04.20.png

Just to bring this full circle, how did you cure the crackling?

Koz

I haven’t cured the crackling completely yet. You defined it perfectly as a blue moon issue, except perhaps it’s not quite as rare as a blue moon. I haven’t narrowed down what conditions are causing the crackling yet. The USB cord wiggling test didn’t produce any crackling. However, the crackling reoccurred on two occasions during recording sessions. During sessions, I periodically pull my headphones out of the mic and place them in the headphone jack of my laptop to listen back. I suspect that perhaps this repeated pulling out and putting back in of the headphones is contributing to the condition that is resulting in the dreaded crackling? That being said, I’m going to bring two pairs of headphones to future sessions and see if the crackling reoccurs.

You defined it perfectly as a blue moon issue

Moon-Phase is a programming/development term of art with an intentionally fuzzy specification of which phase. Blue moons are predictable.

I suspect that perhaps this repeated pulling out and putting back in of the headphones is contributing to the condition that is resulting in the dreaded crackling?

Any self-respecting engineer can concoct a wild-ass set of conditions that can cause any problem.

“The additional current drain by plugging in the headphones causes the USB DC current to rise slightly which causes the oxidation …”

Nope, sorry. Highly unlikely.

When you generate a damaged segment, is the next segment reliably clean? Ever get two in a row?

Koz

I’ll quit the speculation and resume my status as a clueless layman. I haven’t produced two damaged segments in a row. Ever since producing the first crackling segment, I’ve been hyper-vigilant and listen back to my audio every 15 minutes or so. In the two other instances that I detected crackling, I unplugged my mic and started fresh, producing audio without crackling. Any other suggestions for tracking down the source of this issue?

You’ve listened to my audio, you know what the recording space noise levels are like, and you know what equipment I use. I’ve recorded the entirety of this first audio book with these conditions so there is no going back with those clips… Moving forward, should I ditch the Yeti and upgrade, maybe find a new space to record? I know in a perfect world I’d have a sound proof room and professional equipment, but of course, I’m on a budget. What are some tips to improve the quality of my audio and avoid such vigorous editing/mastering?

Very thankful for all the advice and help!
Logan

If you already have an ACX reading, you should go through the corrections and steps for submission to see if they accept it. It’s something like they want a short, cut-down chapter with specific spacing. I only know from people who have submitted, and once they succeed, they tend to vanish as the workload goes through the roof. We’re not company representatives. Sooner or later you have to deal with them.


Studio first. Overcoming room noise and echoes is a major problem and could be impossible. Overcoming echoes can’t be done in post production. A lot of rescues involve soundproofing.

As an experiment, I walked into my soundproofed third bedroom with no microphone and cut a test. That’s my quiet 13" Mac laptop on a bath towel with modest noise reduction and compression. It wasn’t uncomfortable, either. I bet I could read a short book like that…if I could read.

Next adventure is cutting a test with the microphone in an iPhone or iPod. I found an application that will save work in WAV format and then let me have the WAV file afterward. I’ll be recording in my bedroom. I haven’t worked out how to hold them yet. Handling noises are deadly. This is a job for duct tape.

Koz
Screen Shot 2016-07-18 at 20.26.02.png