Freesound Forums

Hi,

I will record some humain voice in various languages. This will basicly be a list of +/- 150 words. The recodings will (obviously) be published here when I got them ready.

Since I got no experience doing recording, I'm seeking advice.

My config:
* A laptop with ubuntu. (vista/Mac is possible, but I rather avoid it)
* a wired singer like microphone (unknow brand, look similar to

[edited : removed]

Sorry, looks like I don't know how to do an A HTMl tag!

Hi,

I will record some humain voice in various languages. This will basicly be a list of +/- 150 words. The recodings will (obviously) be published here when I got them ready.

Since I got no experience doing recording, I'm seeking advice.

My config:
* A laptop with ubuntu. (vista/Mac is possible, but I rather avoid it)
* a wired singer like microphone (unknow brand, look similar to
http://us.123rf.com/400wm/400/400/pierdelune/pierdelune0601/pierdelune060100042/311779-close-up-of-singer-microphone-shallow-dof-focus-on-the-head-isolated-on-white-background-patch-inclu.jpg
used by a friends to do Karaoke.

My Questions:
* Recommandation for recodring setup (avoiding breathing noise, room, ...)
* Which software can I use to record ? To edit the recodring ?
* Should I record all at once, then split it, or better doing a file per word from the start ?
* sound format / quality ?
* typical newbies errors?
* advice to give to the natvie speaker (position, ...)

in two words: I got no clue, any advice is welcome !

Thanks in advance,

PS: I found little ressources fitting my level of knowledge and needs. Audiacy looks like what most people recommand for recording.

Hi mate! Im quite a noob in proper sound softwares and ubuntu. But theres something id sugest and it is to record all in one track, so that way may be a lot faster easier for you to equalise and effect and play with all them at the same time.

Ive got huge experience using shit equipment,so id sugest not to record it too loud by setting the recording volume not too high. That way the sound wont be distorted in the loudest parts. But if you do it too low you may hear a lot of noise when you rise the volume up.

To avoid breathing you can not proyect the air to the mic, but putting the mic in one side of the mouth. The problem is that adding obstacles to a mic makes the sound be more distant and have less detail. A friend of mine used an empty popcorn packet directly over the mic.

But thats not 100% true because that friend was really me.

However, you probably know more than me about that already because your questions dont look like questions by someone who dont know nothing at all!!

Thanks for the tips !

so, you advised me to:
* "Record all in one track for easier post-process" got a lot of sense in fact! I will definitly keep this one.
* "Set the (shity) mic volume a bit lower". Thanks for the tips. I will watch out. (Most of the tips I read say to put it very high to avoid noise. I assume this apply to better quality mic)
* "cover the mic with a light carton box". I will try this one see how its come out. (I assume this is to help for the "p" alike sound)

No, I got no knowledge of the topic.
I just read a bit about it on the net: but most of the tips are like "put the mic at the *good* distance", "*listen* to the room to choose where to put the mic", "buy good material", or the tips are for professionals who already know about the topic and talk about advance technique. so, ... Here I am looking for simple tips !

thanks again,

anybody else got some tips for Software issue / batch recoring or words ?

Some good points so far.

For spoken word intelligibility, it helps to have a very small room and minimal reverb/reflections (can always be added later if need be):

Try to select a room that is small and has some wall coverings. Or find a closet you can make some room in, leave everything that is hanging and separate the hanging clothes evenly to the left and right, and put the mic and stand midway inside the closet (pointing outward) and have the subject read into the closet from about 6-8-10 inches from the mic. (Whichever sounds best and helps eliminate 'p'losives

Make sure the subject is comfortable, and if loud speaking or singing is not required, give the subject a choice of standing vs sitting down. If the subject is comfortable you tend to get better results overall.

Record all your parts in one go, but then have a quick break and do a second one, also all in one go. (Best if they are done on the same day, with all the same parameters). This allows substitutions for errors that might happen. If there is a particular phrase that often gets pronounced badly, tell the subject to repeat that section until you are happy.

If it's individual words or short phrases, breath noises won't a concern. If there will be a breath sound in the middle of a longer phrase, you can edit it out by hand. Don't delete it so the space is removed or it will throw off the timing in an awkward way. In your wav editor, you can highlight it and low the gain. If you have substantial background noise, it will sound wierd if you drop it all out, so you might need to stick in some "room tone" for consistency. Don't record too loud, but try and record at a decent level that still leaves headroom. I suggest you test the speech and record a sample at the loudest you expect it to be. Set the level to where it is well below the maximum. Average -12db is a good average to shoot for, even if it looks very low. This is if you are able to record 24 bit. In such a case, you have plenty of room to get a great sound at a low level so you don't risk overload. If not, try and record louder, still with a safety.

Once the files are recorded, it's the fun part: editing!. Your wav editor will help you split the file into separate pieces and files. Be very conservative with setting it so it gives you more than you need in separate files. Then go an edit them by hand, taking out unnecessary silence at the beginning and end. If you chop out too much it will sound bad, so it's always better to leave more than you need in a file than not have what you do, such as initial attacks of words or ending bits (think of the last "s" in "potatoes". It happens well after most of the the word is finished and tails off very quiet. But you need it or you've only got a potato).

You shouldn't need much processing, if any. Normalize if the levels are way too low, and probably hipass. You can look this up or ask when the time comes.

Thanks for all those practical tips!

I will upload the files soon.