I finally bit the bullet and shelled out for a decent audio system. I got a Shure SM58 vocal mic (with foam wind shield) and hooked it up to the Onyx Mackie blackjack 2x2 audio interface. I've got the mic sitting on a cheap boom arm (think old school desk lamp with spring arm) so the mic can be easily positioned in front of me (i.e. a few inches from my mouth).
The results have been extremely gratifying so far. I'm still exploring the system but I have produced several video podcasts for classes now, ranging from 15 to 30 minutes in length. The audio is clear with next to no background noise (despite being in a room with a computer fan and air con).
I'm still finding that I need to bump up the audio levels in post processing to get something that sounds ok over the built-in external speakers (but for headphone listening that isn't required). I'm not sure how many people will be using external speakers rather than headphones but it isn't too tricky to increase the gain (I've just been using a normalisation filter in bbflashback to do the job).
The only downside is that it would be a pain to move the setup, so if I wanted to use it in a lecture I would be carting quite a bit of stuff down with me. Everything would pack down reasonably small though, so at a pinch I *could* take it all to a lecture theatre if I had to. It would not be convenient though.
Monday, 31 March 2014
Thursday, 20 March 2014
Video conversion (so you need a smaller file...)
When recording lectures I use one of two tools:
1) The automated lecture recording software provided by the university
2) Bbflashback on my surface
The first option has the advantage everything is done for you and an email turns up in your inbox with links to whatever you might want (a low quality, high quality and audio only recording). The disadvantage is that even the low quality recording is pretty large, at around 100Mb for a 50 minute lecture. For students based on campus the large file size isn't a problem. For extramural students it is a different story, as their internet connection isn't necessarily reliable or fast.
The second option works well if you want a smaller recording and gives much more control (but I am having problems getting decent noise-free audio). As bbflashback records only changes to the screen (rather than multiple frames a second) you can end up with a much smaller file size (e.g. 15 Mb for a 50 minute lecture). This assumes you export with a very low frame rate and relatively low quality (but it makes for a perfectly viewable recording). It also assumes you are using something like power point or writing on the tablet, rather tha playing video footage.
For some courses I use both, using the automated recording as a backup. This was helpful this week when I accidentally hit the "discard" rather than "save" button after recording a bbflashback lecture.
All was not lost because of the automated recording but I really wanted to convert it from a 100Mb file to something smaller.
After a bit of research I settled on a video conversion tool called Handbrake: http://handbrake.fr/ (open source and available for free).
This read my mp4 recording in with no issues and then I set about to tweak settings to see if I could export it as a smaller file. Almost everything I tried in terms of optimisation resulted in an approximately 60Mb file. Tweaking different video settings didn't make much difference (I tried scaling the video size to half the original size, going to greyscale, using higher compression, longer encoding time, lower quality and nothing brought the size down by more than a Mb or two).
In the end I concluded that if altering the video didn't change things much then maybe I should play with the audio instead. After all, a 3 minute song can easily be 3Mb, so an hour of audio could account for a lot of that 60Mb file size. Turns out this hunch was right on the money. By dropping down from 128bps to 64bps I was able to reduce the file size to 33Mb.
The lesson is if you want to reduce your video recording file size, try downgrading the audio quality, you may be pleasantly surprised how much of a space saving that gives you (with no loss of intelligibility for spoken word)
1) The automated lecture recording software provided by the university
2) Bbflashback on my surface
The first option has the advantage everything is done for you and an email turns up in your inbox with links to whatever you might want (a low quality, high quality and audio only recording). The disadvantage is that even the low quality recording is pretty large, at around 100Mb for a 50 minute lecture. For students based on campus the large file size isn't a problem. For extramural students it is a different story, as their internet connection isn't necessarily reliable or fast.
The second option works well if you want a smaller recording and gives much more control (but I am having problems getting decent noise-free audio). As bbflashback records only changes to the screen (rather than multiple frames a second) you can end up with a much smaller file size (e.g. 15 Mb for a 50 minute lecture). This assumes you export with a very low frame rate and relatively low quality (but it makes for a perfectly viewable recording). It also assumes you are using something like power point or writing on the tablet, rather tha playing video footage.
For some courses I use both, using the automated recording as a backup. This was helpful this week when I accidentally hit the "discard" rather than "save" button after recording a bbflashback lecture.
All was not lost because of the automated recording but I really wanted to convert it from a 100Mb file to something smaller.
After a bit of research I settled on a video conversion tool called Handbrake: http://handbrake.fr/ (open source and available for free).
This read my mp4 recording in with no issues and then I set about to tweak settings to see if I could export it as a smaller file. Almost everything I tried in terms of optimisation resulted in an approximately 60Mb file. Tweaking different video settings didn't make much difference (I tried scaling the video size to half the original size, going to greyscale, using higher compression, longer encoding time, lower quality and nothing brought the size down by more than a Mb or two).
In the end I concluded that if altering the video didn't change things much then maybe I should play with the audio instead. After all, a 3 minute song can easily be 3Mb, so an hour of audio could account for a lot of that 60Mb file size. Turns out this hunch was right on the money. By dropping down from 128bps to 64bps I was able to reduce the file size to 33Mb.
The lesson is if you want to reduce your video recording file size, try downgrading the audio quality, you may be pleasantly surprised how much of a space saving that gives you (with no loss of intelligibility for spoken word)
Wednesday, 19 March 2014
In search of better screencasting audio
I have been using my surface pro for screen casting every now and then and am slightly frustrated with a few things. I'm after clear video, great audio and a small file size. Not much to ask :-)
On one course I use bb flashback to record live lectures (as opposed to our usual automated lecture recording system provided by the university). The audio gives me speech that is perfectly understandable but there is quite a bit of background noise and it isn't that pleasant to listen to. I noticed that even when recording in my office I still get a fair bit of background noise when using the built-in microphone. I suspect it is picking up my office air conditioning and it is definitely picking up the surface fan when it kicks in. This is a problem I want to solve. Initially I thought it would be as simple as shelling out $20-$50 for a decent external microphone. Turns out it isn't that simple.
I've been doing some reading around audio recording and after trialling a bunch of different microphones and audio interfaces have come to the following conclusions:
a) to avoid electronic noise, use usb
b) to avoid background noise from the environment use a good quality directional dynamic microphone (as opposed to a condenser)
c) microphones may need an amplifier to boost the signal
d) if you want really nice audio, expect to shell out a lot for it (e.g. around $500 NZ)
I'll deal with each conclusion in turn.
Cheap external computer microphones (or headset microphones) will often plug into the 3.5mm jack on a laptop. With the surface pro I have found that every microphone that I've plugged into a 3.5mm jack has had issues with noise. My brother who is an electrical engineer (and audiophile) told me that using the 3.5mm jack and onboard circuits for sound processing is likely to result in noise. The included soundcard on portable computers isn't going to be ideal for processing high quality audio. It is better to ditch the 3.5mm jack and get your audio in via the usb connection (so that the signal processing can be done externally). There are a number of well known microphone manufacturers who produce microphones that plug directly into a usb port. Here are a list of the ones I considered after extensive research: the Rode podcaster, the Blue Snowball, the Blue yeti, the Studio projects LSM, the Samson Go-mic and the Audio-Technica ATR2100 USB. You can find youtube reviews on all of these but the most useful site I found was Recording Hacks: http://recordinghacks.com/
All these microphones come highly recommending for podcasting.
Microphones fall into two main categories, condenser microphones (which use a capacitor) and dynamic microphones (which use a magnet and coil). Condenser microphones tend to be much more sensitive and also need to be powered. I tried out a few condenser mircophones (e.g. the Blue Snowball and Blue Yeti) and while the audio was very clear and electronic noise wasn't an issue, background noise was. Despite these microphones having cardiod pickup patterns (which means in theory they only pick up noise in front of them, rather than from all around) condensers are so sensitive that I as picking up noises from behind the microphone at the very far end of the shop I was in. Condensers would be great in a studio environment with sound proofing where background noise wasn't an issue. Since I have air conditioning on in my office, plus a computer fan and the sound of me writing on my surface with a stylus, condensers weren't going to work. By contrast the dynamic cardiod microphones I tried were MUCH better at picking up just my voice and no background noise. They are less sensitive so you have to position your mouth a lot closer to the microphone but this is a small price to pay for less noise. The Rode podcaster did a lot better job of isolating my voice, even though I was testing it in a relatively noisy shop.
Unfortunately the Rode podcaster didn't give me very high input levels so I had to post process the audio recordings and massively scale the gain. It appears that this is a common complaint with usb microphones and that it is often necessarily to increase the amplitude of the signal. While you can do this in software, increasing the signal via post processing can also introduce audible background noise (that wasn't noticeable until you increase the amplitude). A better solution is to use a pre-amp (pre-amplifier) to increase the signal.
If you are going the route of using a pre-amp, then you need an audio-interface, which does the job of both amplifying the microphone's signal and converting the signal to a digital format so that it can go in via usb. The nice thing about going with this route is you are no longer limited to usb microphones, you can use professional audio microphones which make use of the standard XLR connectors present on just about all mics used in the recording industry. I tried out The Shure SM57 instrument mic (an industry standard) with my brother's audio interface (a TASCAM audio interface ) and got fantastic quality. This may well be the way to go.
In summary, to get really good audio you will probably need a good audio interface (for around $200-$300) plus a good dynamic microphone (for around $150-$250) plus stand and cable.
I've been looking at the Shure SM58 or SM48 for the microphone and the Onyx Mackie blackjack as the interface.
My last hope at a cheaper solution is the Audio-technica AT2100USB microphone, which is currently out of stock, so I haven't been able to try it yet.
On one course I use bb flashback to record live lectures (as opposed to our usual automated lecture recording system provided by the university). The audio gives me speech that is perfectly understandable but there is quite a bit of background noise and it isn't that pleasant to listen to. I noticed that even when recording in my office I still get a fair bit of background noise when using the built-in microphone. I suspect it is picking up my office air conditioning and it is definitely picking up the surface fan when it kicks in. This is a problem I want to solve. Initially I thought it would be as simple as shelling out $20-$50 for a decent external microphone. Turns out it isn't that simple.
I've been doing some reading around audio recording and after trialling a bunch of different microphones and audio interfaces have come to the following conclusions:
a) to avoid electronic noise, use usb
b) to avoid background noise from the environment use a good quality directional dynamic microphone (as opposed to a condenser)
c) microphones may need an amplifier to boost the signal
d) if you want really nice audio, expect to shell out a lot for it (e.g. around $500 NZ)
I'll deal with each conclusion in turn.
Cheap external computer microphones (or headset microphones) will often plug into the 3.5mm jack on a laptop. With the surface pro I have found that every microphone that I've plugged into a 3.5mm jack has had issues with noise. My brother who is an electrical engineer (and audiophile) told me that using the 3.5mm jack and onboard circuits for sound processing is likely to result in noise. The included soundcard on portable computers isn't going to be ideal for processing high quality audio. It is better to ditch the 3.5mm jack and get your audio in via the usb connection (so that the signal processing can be done externally). There are a number of well known microphone manufacturers who produce microphones that plug directly into a usb port. Here are a list of the ones I considered after extensive research: the Rode podcaster, the Blue Snowball, the Blue yeti, the Studio projects LSM, the Samson Go-mic and the Audio-Technica ATR2100 USB. You can find youtube reviews on all of these but the most useful site I found was Recording Hacks: http://recordinghacks.com/
All these microphones come highly recommending for podcasting.
Microphones fall into two main categories, condenser microphones (which use a capacitor) and dynamic microphones (which use a magnet and coil). Condenser microphones tend to be much more sensitive and also need to be powered. I tried out a few condenser mircophones (e.g. the Blue Snowball and Blue Yeti) and while the audio was very clear and electronic noise wasn't an issue, background noise was. Despite these microphones having cardiod pickup patterns (which means in theory they only pick up noise in front of them, rather than from all around) condensers are so sensitive that I as picking up noises from behind the microphone at the very far end of the shop I was in. Condensers would be great in a studio environment with sound proofing where background noise wasn't an issue. Since I have air conditioning on in my office, plus a computer fan and the sound of me writing on my surface with a stylus, condensers weren't going to work. By contrast the dynamic cardiod microphones I tried were MUCH better at picking up just my voice and no background noise. They are less sensitive so you have to position your mouth a lot closer to the microphone but this is a small price to pay for less noise. The Rode podcaster did a lot better job of isolating my voice, even though I was testing it in a relatively noisy shop.
Unfortunately the Rode podcaster didn't give me very high input levels so I had to post process the audio recordings and massively scale the gain. It appears that this is a common complaint with usb microphones and that it is often necessarily to increase the amplitude of the signal. While you can do this in software, increasing the signal via post processing can also introduce audible background noise (that wasn't noticeable until you increase the amplitude). A better solution is to use a pre-amp (pre-amplifier) to increase the signal.
If you are going the route of using a pre-amp, then you need an audio-interface, which does the job of both amplifying the microphone's signal and converting the signal to a digital format so that it can go in via usb. The nice thing about going with this route is you are no longer limited to usb microphones, you can use professional audio microphones which make use of the standard XLR connectors present on just about all mics used in the recording industry. I tried out The Shure SM57 instrument mic (an industry standard) with my brother's audio interface (a TASCAM audio interface ) and got fantastic quality. This may well be the way to go.
In summary, to get really good audio you will probably need a good audio interface (for around $200-$300) plus a good dynamic microphone (for around $150-$250) plus stand and cable.
I've been looking at the Shure SM58 or SM48 for the microphone and the Onyx Mackie blackjack as the interface.
My last hope at a cheaper solution is the Audio-technica AT2100USB microphone, which is currently out of stock, so I haven't been able to try it yet.
Subscribe to:
Posts (Atom)