fixing my mistakes

Hello, I am sure this has been covered but i have searched for this in every way I could think to ask it and didnt find what I was looking for so I am going to ask.

OK so I record audio books (working on my first) I have figured out almost everything except in my editing. when my author gets back with me about a miss-pronunciation I go and record my fix, and insert it in the file. Here is where my problem starts I can never make it sound the same. I know there has to be a way to match the corrected audio to the rest but for the life of me I cant get it to blend in. here is an example of my problem :blush:



Thank you for any help you can give me and also for all the help I have been able to get using your fourm

Very few people can get a word to slide in gracefully. Re-record the sentence or segment of narration that has natural breaks so the difference won’t be as obvious.

We can’t tell much from two seconds, but it sounds like you changed the face/microphone distance. That’s rough to manage in post production. I have appeared briefly as two different people by just changing the microphone distance. It’s a very serious sound change.

Again, two seconds isn’t much, but it’s a common new-user error to think they have to “live record” a correction like that. Make a protection copy of the “broken” work and announce the correction by just pressing record. Audacity should produce a second track with the correction on it. Tune which track to listen to with the SOLO and MUTE buttons.

Then use the Time Shift Tool (two sideways black arrows) to slide the corrected word into position, and the envelope tool to mute the old word at the right place. Audacity will smash both tracks together when you export.

http://manual.audacityteam.org/man/envelope_tool.html

This lets you add or subtract time, effects, and volume to the correction without affecting the main work. I would have increased the time between the existing and the correction, for example, which is relatively easy when they’re two separate tracks.

And all that is why it’s best to re-record the whole sentence or paragraph. It’s far easier to hide minor changes like that.

Have you run into the ACX quality control services yet? We warn people not to record the Whole Book only to discover ACX won’t accept your sound quality.

Koz

Yes some of your other posts of been a huge lifesaver as far as ACX goes I usually do record a section or symptoms like you suggest but I still can’t ever get it to sound the same I I don’t know if I speak differently or what it is exactly I do but it never sounds the same and it frustrates me because I’ve had to re-read entire chapters because I can’t get it to blend close enough and then they find more errors in my new recordings but I fix the old ones it’s extremely frustrating to me and it almost has me wanting to just give up I can post a better sample in just a minute

they find more errors in my new recordings but I fix the old ones

It’s not the easy, fluffy job many people think it is.

You can post up to twenty seconds of mono (one blue wave) work on the forum.

Describe how you’re recording your voice. It’s not unusual for the computer’s built-in corrections and tools to be “helping you” when that’s the last thing you need. Computers are business machines not generic tools they used to be. You spend a lot of time turning off built-in corrections that may work OK in a conference to the head office, but sound terrible in an audiobook or podcast.

http://manual.audacityteam.org/man/faq_recording_troubleshooting.html#enhancements

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

Koz

here is a reread though a section didnt do much it so you can advise me master lol :smiley:

I have a headset microphone that I am currently using, audacity is my program, i use the nyqust script you and steve recomend nd the limiter

I didnt expect it to be a real fluffy job it was something I have had a desire to do.

I don’t see a sound clip.

Koz

There is a much more exotic way to do this. Use the musical overdubbing techniques (record all the instruments in a song one at a time).

Listen to the original track from the beginning of the segment and read right along with it so you have exactly the same rhythm and volume and announce the correction when you get there. Much better than trying to announce the correction cold.

That could take more effort to set up and get used to than may be worth, but that is another way. That’s a cousin to “smash recording” which is an older technique Audacity can’t do. Play the older work and smash record on the tape recorder when you get to the fluff.

Koz

sorry I guess my file was to big but here it is.

Here is where my problem starts I can never make it sound the same.

This won’t help you…

I saw a clip on TV of an author recording his audiobook. There was a producer monitoring on the other side of the glass and every once in awhile, the producer would stop him and ask him to start-over (I think at the beginning of a paragraph). And sometimes, he would give him “hints” on pronunciation, etc. Everything was corrected immediately so it was the same setup and the same “voice”. The “tape” was never stopped. The paragraphs/sections were numbered and I believe the author would announce each paragraph/section so everything was easy to keep track of and it was easy for the producer to tell him were to re-start. And, I’m pretty sure the producer was keeping notes on his copy.

This particular author had an advantage in that he was a radio/TV professional so he was used to being behind the mic every day.

There’s a 2MB limit to forum posting. That’s either 20 seconds of mono (one blue wave) or 10 seconds of stereo. If you need to post longer WAV samples, you have to do it using third parties such as DropBox


The longer posting is very highly compressed and processed. It’s not a “clean” recording and that’s one reason different passes for different corrections don’t match.

There’s a quality of talking into and out of a wine glass typical of excessive noise reduction, or Windows conference processing.

Listen carefully on headphones just after about 13 seconds. You take a breath between spoken phrases, but the sound is strangled like a bad cellphone.

That may pass automated quality control, but it will never pass Human QC.

See if anything here helps.

http://manual.audacityteam.org/man/faq_recording_troubleshooting.html#enhancements

If that sample was more or less plain recording, then Windows is poking its fingers in it.

Koz

thank you for your help I havent gotten close to perfect yet but much closer then I was

Post a sound clip according to that formula after you beat Windows into submission.

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

Sometimes we can identify problems straight-away and save you a lot of work.

It’s not kidding. Make a simple recording and do not do anything to it before you post it.

I like the voice. We just have to get it recorded correctly.

Koz

here you go, this is completely unedited.

The voice-signal on that is way too low …


The voice-signal is only ~22dB above the constant noise-floor. You should be aiming for ~50dB above noise-floor, (then you don’t need to use lots of noise-reduction, which makes the voice sound computery).

You may have a physical mic-gain knob on your hardware somewhere,
or a virtual gain-slider control in the software, to increase the voice-signal, e.g. …

http://manual.audacityteam.org/man/tutorial_making_a_test_recording.html

Why is to be determined, but yes, the “rain in the trees” hiss noise behind you is way too loud. And/Or, your voice is way too low.

We’re 2/3 there. I applied the audiobook tools and got everything to pass including voice quality…except noise.

Before we get into the weeds about microphone internal noises, etc, who is that talking behind you? There are two voices in there. You super can’t do that. We can’t split two voices apart in post production and delete only one. Once they’re mixed together, that’s the show.

The good news is you’re missing all the processing and built-in craziness and distortion that the machine was applying. The bad news is the processing was covering up some problems.

As above, make sure you have the volume controls in your microphone, Windows and Audacity turned up. After that, announce a test and watch the blue waves and Audacity sound meter. Slowly get louder and closer to the microphone and watch Audacity. I expect you to be about a power fist away from your pop filter or a Hawaiian Shaka away from your microphone.

And be able to get something like this.

What microphone do you have? Getting this spacing is rough if you have a table mounted microphone.

“Bend down until your chin is on the table…”

Just kidding.

Koz

My fault. You did say what you were using.

I have a headset microphone that I am currently using

Describe it. Model numbers?

Koz

See? I can force it to pass ACX AudioBook compliance (see first three readings and sentence 2/3 down)
Screen Shot 2017-11-18 at 16.02.37.png
But I had to go through noise processing twice to get there and your voice is starting to sound hollow and wine-glassy.

If you play that in Audacity, note how much louder it is than your raw clip. You should get closer to that volume when you record.

Koz

I have two headset type microphones.

One is a Logitech “gamer” type full USB headset and it’s pretty awful. It works, but the sound is constricted, honky and low volume. I put it in the garage. The other is a head-mounted theatrical microphone and it’s a lot more successful. It takes a small sound mixer and USB interface to run it. That’s the one I used for my Skype Podcast test.

http://kozco.com/tech/audacity/clips/DenisePodcastMoreHeavilyCut3.mp3

If you do it right, those can eliminate a lot of the problems of a noisy or echoy room, but you do have to use them just right. It’s easy to get P popping and breath wind noise.

Koz