Too much natural reverb

[mbrenneman]

Audacity=1.2.6
Recorder=Roland Edirol R-09
Recorder settings= I've tried gain on high and low, low-cut on and off
Settings=I've tried recording in WAV at 44.1 and 48KHZ, 16 bit and 24 bit; encoding to MP3 at 48 to 160 bit rate

Problem: I work in a public library, and, as most public libraries, it's really noisy. The only quiet place to record in my building is a conference room with one glass wall, one brick wall, and 2 regular smoothish walls. Furnishings are a shiny table and some upholstered chairs, some silk plants, carpet, and that's all. I'm recording voice for a podcast, so I take the portable recorder into the conference room and cover the table with coats I've collected from my colleagues. I position my guy sitting facing the brick wall with the recorder between him and the bricks, with the silk plants positioned around him to the sides and back and the coat-covered table behind the plants. I would have to cover the walls to get less bounce in there.

When the recording is playing in WAV format, it sounds fine, and I even like the sound of the room. When it's compressed to 64-bit MP3, it still sounds acceptable on speakers. The problem is that since it's a podcast, I think more people will listen on earbuds and THAT'S when it sounds horrible. It sounds barely acceptable on earbuds at higher bit rates, but I'd like to keep file sizes small since it's just speech.

I've tried some Audacity effects just to try to get the sound clean in general, but I can't hear much difference after compression to MP3:
Dynamic range compression at various settings: threshold -12 to -15; ratio at 2:1 to 3:1; attack at .1 to .2
Low pass filter at 10,000
High pass filter at 100

I can't guess what I should try with EQ. Wouldn't the reverb be at the same frequencies as the guy's voice?

Another thing I'm guessing is that maybe it's something about the encoder, so the next thing I'll try is different software for conversion to MP3.

If anybody is kind enough to listen on cheap headphones, my first two published attempts are Historic Knoxville News #1 and #2 at http://blogs.knoxlib.org/main/podcasts/. The first one was published at 112 bit rate with no filter effects, the second at 160 with some attempt at filtering. The part to hear is when Robby is reading the news, so fast-forward to about a minute.

I'm working on the third podcast now. Thanks for any tips!


--
Melissa Brenneman
Reference Librarian
Knox County Public Library
500 West Church Avenue
Knoxville, TN 37902
(865) 215-8723

[mbrenneman]

I'm replying to my own post because I just now noticed that Koz had answered a previous post of mine in another topic. I'll repeat his excellent post here and then reply to it below.

Koz said:


Gotta love those moonshiners.

OK, now we know what we're dealing with. The music works and man's fine, but the woman's voice is, at times, harsh and peaky. It's difficult to believe that's happening in the digital world, so there must be something happening to the microphone or the microphone amplifier. We can also probably assume that the woman doesn't sound like that normally.

Can you point me back to the original posting, or just tell me what kind of microphone or equipment was used? That distortion will be very nearly impossible to get rid of in post production without destroying the quality of the voice, so we have to clear it up in the live capture stage. The other obvious question would be, were the man and woman using the same microphone?

Oh, and without offending anybody's sensibilities, the live capture could happen in the bedroom. A carpeted bedroom has far less reverberation and echo than normal rooms or certainly far less than the average office. I've been known to spread all the recording equipment out on the bed--including the microphone--and go with it.

That's how I recorded the Left-Right clip from here. Doesn't sound like I'm recording on a Queen size BeautyRest on a brass bed, does it?

http://www.kozco.com/tech/LRMonoPhaset4.wav

Koz


Hi, Koz, I'm sorry that I just now read your March 2 post in the forum " If you've made a project..." where you gave me some tips. I thought the forum would e-mail me an alert if I got an answer. duh. I must have skipped checkmarking that box.

So, you did suggest setting up the room for less bounce, and that's something I'll really have to think about. Don't know whether it would be better to record hunkered down within the book stacks before the library opens, maybe that's what I need to do.

During the interview part, I'm sitting next to the interviewee with the recorder on a table (spread with coats and/or cardboard boxes) more in front of the history expert, so I'm sitting to the side. I'm (unfortunately) very sibilant, but I never saw the recorder peak when I spoke.

When I'm doing the intro, transition and bumper parts, I'm usually less uptight and that probably cuts down on the sibilance a lot, but also sometimes I can just hold the recorder in my hand and that might be helping, also. I've asked for some grant money for a boom stand and lavalier mic for the recorder to make it easer to record in other situations, too.

Thanks again, Koz...

--Melissa

[steve]

Filtering the (voice) audio (100 Hz / 10 kHz) is a good idea and will help minimise unwanted noise without loosing any significant sound quality.

You can also make your mp3's much smaller by encoding as mono. don't think you can do that (at least not easily) in Audacity, so you will need to export from Audacity as a WAV file, then use an external encoder such as LameDrop to convert to mp3. 80 kbps mono will be smaller, and better quality than 120 kbps stereo.

[mbrenneman]

Here's the result of a test on my home computer this weekend. Yesterday at the library I had exported my project in WAV to try encoding with other software, as stevethefiddle suggested. Took it home and opened it on my 6-yr-old Sony Vaio PC in Screenblast Sound Forge 1.0, which I've never played with before (I thought Audacity and Cakewalk were less intimidating, LOL). I exported to MP3 at 64 Kbps. Put it on my Creative Zen player and listened through earbuds.

Result: WOW! What An Improvement !!!! It sounds almost like it's done by somebody who knew a little something about what they were doing. Hmmm. So, lesson learned: the encoder can make an enormous difference. And I still need to work on reducing bounce in the recording process.

In Screenblast I found a tool called Bit Depth Converter and I set it to 16-bit, Highpass Triangular dither, High-pass contour noise shaping. I don't even know what those are, but thought it might be worth a try. After that took effect I couldn't really detect much difference in the resulting mp3. I guess for my third podcast I'll keep it simple on the filter effects and take it home to encode with Screenblast. I'm still open to any more suggestions, but thanks to Koz and stevethefiddle.

--Melissa

[steve]

A couple of tips about reducing the natural room reverb;

1) Placing the microphone closer to the person talking will increase the amount of direct sound. so giving a relative reduction in room ambiance. When the microphone gets very close there are a couple of things to watch out for though:

a) you must avoid blowing on the microphone as this will cause a lot of pops and booms whenever the speaker says a word beginning with the letters "P" or "B". Ways to avoid this are to make sure that the microphone is not directly in front of the persons mouth - ideally placed at a hight level with the end of the persons nose and pointing the microphone directly at them, or failing that, below the persons mouth. Also a Pop-shield helps a lot, you can make a simple one out of coat hanger and a pair of ladies tights (not fishnet), or purchase one quite cheaply. (see below for picture). There can be some benefit from the foam "windshields", but the first type are much more effective.

b) when the microphone gets very close to the person (within a couple of inches) there will be a noticeable emphasis of the bass frequencies. You can compensate for this by using Equalisation to lower the amount of bass (bass roll-off).

2) If you can get hold of coat stand, you could try placing this behind the microphone (between the microphone and the wall) with a few coats on it. This may help reduce the amount of sound picked up by the rear of the microphone (reflections from the brick wall)


Pop shield between mouth and microphone to prevent air blasts hitting the microphone.


A foam windshield - better than nothing but not as good as a proper pop shield.