Skip to main content ↑ ↓

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: FM Broadcast: 44 kHz or 48 kHz (Read 12921 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

FM Broadcast: 44 kHz or 48 kHz

Hello,

Anyone know wich Sampling Rate will be better for recording analog radio, FM to WAV.

http://en.wikipedia.org/wiki/FM_broadcasting

This article have some related information, but I cant understand it :\

FM Broadcast: 44 kHz or 48 kHz

Reply #1 –
Since analog FM is restricted to less than 15 kHz, 48 kHz would be overkill. You will probably also want to set your lowpass to reject the 19 kHz pilot tone.

FM Broadcast: 44 kHz or 48 kHz

Reply #2 –
You should choose based on your recording equipment, e.g. if it's an AC97-based audio device, then use 48kHz for recording to avoid device-dependant resampling. If needed, apply higher quality postprocessing (resampling, lowpass, etc.) afterwards.

FM Broadcast: 44 kHz or 48 kHz

Reply #3 –
Lowpass on FM is so low it doesn't matter.  Just use whatever your equipment works best at (probably 44.1 or 48khz) and then resample later if needed.

FM Broadcast: 44 kHz or 48 kHz

Reply #4 –
Quote from: ZinCh on
This article have some related information, but I cant understand it :\

Ask away what exactly you didn't understand.

As has been pointed out in most cases you don't actually have the liberty to adjust the sampling rate. Record at whatever the rate is set to. (Test with tone sweep at various rates.)

Don't expect to improve the quality much though. The actual quality bottleneck is usually the receiving antenna. That's where you should look for improvements.

FM Broadcast: 44 kHz or 48 kHz

Reply #5 –
Quote from: j7n on
Don't expect to improve the quality much though. The actual quality bottleneck is usually the receiving antenna. That's where you should look for improvements.

How is the receiving antenna a bottleneck?

FM Broadcast: 44 kHz or 48 kHz

Reply #6 –
Well, if there is no real antenna installed and you're receiving using just a piece of wire, then even the most expensive soundcard won't make a difference in quality. As it was pointed out several times already, there is little perceivable difference between various soundcards today.

FM Broadcast: 44 kHz or 48 kHz

Reply #7 –
Quote from: j7n on
Well, if there is no real antenna installed and you're receiving using just a piece of wire, then even the most expensive soundcard won't make a difference in quality. As it was pointed out several times already, there is little perceivable difference between various soundcards today.

I have gotten excellent reception with just a piece of wire. Of course, I was less than a mile from twelve powerful FM stations.

FM Broadcast: 44 kHz or 48 kHz

Reply #8 –
An older brother of mine used to live across the road from the local FM repeater and managed to blow up the input stage of a very nice tuner and physically bend the needle on the moving-coil signal strength meter with a single length of wire 18" long plugged in as an antenna.

Good antennas are great and their value should never be underestimated. A nice multi-element is going to provide a cleaner signal by virtue of having a relatively narrow field of view and a very good front/back ratio when compared to the 360 degree field of view of a straight piece of wire. This reduces multi-path effect which is largely compensated for in DAB radio with error-correction, but is hard to get rid of in an analogue tuner.

Maybe the best idea is to buy a good (but not necessarily expensive) multi-element antenna and then shove it through an in-line attenuator at the tuner end if it proves necessary. It wipes the floor with anything else if all of the stations you require come from the same transmitter, which here in the UK, they usually do. Especially if you live out of town because almost everything you can pick up comes from your local regional repeater anyway.

Cheers, Slipstreem. 

FM Broadcast: 44 kHz or 48 kHz

Reply #9 –
thanks for replies, I will check wich sampling rate support my soundcard ( SigmaTel STAC 92XX C-Major HD Audio )

Quote from: pdq on
You will probably also want to set your lowpass to reject the 19 kHz pilot tone.


How I can do that? Its some encoder setting? Any other special encoder setting recommendations?

FM Broadcast: 44 kHz or 48 kHz

Reply #10 –
I just record at 32khz and forget about it.

FM Broadcast: 44 kHz or 48 kHz

Reply #11 –
As daft as it may sound, it's possible that using LAME in VBR at -V7 could give quite acceptable results. It's specifically tailored to have a cut-off frequency  of around 14.5 to 15kHz as well as using Joint Stereo and resampling to 32kHz by default. There isn't anything worth recovering from an FM Stereo radio signal above 15kHz, so you're not going to miss anything of any value in that respect. You'll actually be helping the encoder to work more efficiently by filtering out unwanted noise from the signal.

I would expect that to occupy a bitrate of ~100Kbps and give the best quality available at that given bitrate from LAME. It's certainly worth trying.

Cheers, Slipstreem. 

FM Broadcast: 44 kHz or 48 kHz

Reply #13 –
I am using stable LAME release:

Code: [Select]
  Filter options:
  --lowpass <freq>        frequency(kHz), lowpass filter cutoff above freq
  --lowpass-width <freq>  frequency(kHz) - default 15% of lowpass freq
  --highpass <freq>       frequency(kHz), highpass filter cutoff below freq
  --highpass-width <freq> frequency(kHz) - default 15% of highpass freq
  --resample <sfreq>  sampling frequency of output file(kHz)- default=automatic


So I need to use just " --lowpass 19 " in LAME?

Code: [Select]
lame --lowpass 19 input.wav output.mp3 -V [...]
LAME 3.98 32bits (http://www.mp3dev.org/)
CPU features: MMX (ASM used), SSE (ASM used), SSE2
Using polyphase lowpass filter, transition band: 18774 Hz - 19355 Hz
[...]


something like this?

FM Broadcast: 44 kHz or 48 kHz

Reply #14 –
Quote
How I can do that? Its some encoder setting? Any other special encoder setting recommendations?

If you're gonna use lossy compression, good encoders (LAME, OggEnc2) allow you to set the lowpass frequency. Uncompressed audio can be lowpassed very well using Cool Edit Pro.


Quote from: Slipstreem on
As daft as it may sound, it's possible that using LAME in VBR at -V7 could give quite acceptable results. I would expect that to occupy a bitrate of ~100Kbps.

If the material is worth recording, then it is also deserves a reasonable bitrate. I wouldn't choose 100 kBit/s for any stereo material today. It's a myth that radio quality "equals" 96, or 100 kilobits.

You'll need to set lowpass to 16 kHz. 19000 Hz is the center frequency of that signal, but there are also 2k wide sidebands of unknown purpose present.

FM Broadcast: 44 kHz or 48 kHz

Reply #15 –
Quote from: ZinCh on
So I need to use just " --lowpass 19 " in LAME?

Code: [Select]
lame --lowpass 19 input.wav output.mp3 -V [...]
LAME 3.98 32bits (http://www.mp3dev.org/)
CPU features: MMX (ASM used), SSE (ASM used), SSE2
Using polyphase lowpass filter, transition band: 18774 Hz - 19355 Hz
[...]


something like this?

Yes, something like that, but 19 kHz is still in the pass band so I would go with something lower, such as 15 or 16 kHz. You will still not be filtering out any useful signal, only noise.

FM Broadcast: 44 kHz or 48 kHz

Reply #16 –
Possibly confusing matters still more, i'd like to add that using 32khz sampling rate may not be as good idea as it may seem at first.
Reasons:
*) due to the lower sampling rate, the preecho effect is higher.
*) encoders are usually tuned at 44Khz
*) which resampling would you use? (soundcard, encoder's default one, 3rd party one?)
*) And..well... for playback... it's gonna be resampled too, probably.


My recommendations are as follow:
*)As suggested above, first try to get a clean signal. Get an aerial antenna if you can, so that the reception is good.
*)Capture to your pc at 44 or 48khz (since you mention your soundcard is an HD one, then it is not an AC-97, so it should be ok to sample at 44, but it's up to your choice. there's not much to gain or lose)
*) About the pilot tone, there's the possibility that the receptor already filters it. If not, you can filter it out when encoding (if using lame) with --lowpass 16
*) about the quality/bitrate , i would opt somewhere between -V 5 and -V 3. There shouldn't be bitrate bloat, because the contents over 16Khz should be little to none, so even using a higher quality shouldn't increase the bitrate too much.

Of course, these are just suggestions, but should be good enough.

[edit: reduced lowpass from 18 to 16 from comments of previous posts)

FM Broadcast: 44 kHz or 48 kHz

Reply #17 –
Quote from: j7n on
If the material is worth recording, then it is also deserves a reasonable bitrate. I wouldn't choose 100 kBit/s for any stereo material today. It's a myth that radio quality "equals" 96, or 100 kilobits.
In what way is it a myth? How many bits per second on average do you need to LAME encode an audio signal with a 15kHz upper cut-off and limited channel separation? Assuming a clean recording, isn't it going to be far less than recording from CD sources with a 20kHz upper cut-off and potentially infinite channel separation?

Unless LAME makes worse decisions regarding dynamic bitrate control at the lower -V settings, is there any real reason why the output should sound noticeably worse when LAME happens to be coincidentally custom-tuned to the source material by using exactly the right parameters in some respects at -V7?

I'm not being deliberately argumentative, but I'd be happier if I could hear some samples using this approach and others to see just how much difference there really is in practice.

Cheers, Slipstreem. 

PS I'm asking questions, not making statements.
 

FM Broadcast: 44 kHz or 48 kHz

Reply #18 –
Quote from: Slipstreem on
Unless LAME makes worse decisions regarding dynamic bitrate control at the lower -V settings, is there any real reason why the output should sound noticeably worse when LAME happens to be coincidentally custom-tuned to the source material by using exactly the right parameters in some respects at -V7?

The V parameter, as you know, defines quality, not bitrate.
That said, one thing is the desired goal, and another is the implementation of that goal.
Concretely, i mean that the setting is not only changing the parameters on sampling rate and lowpass, but also making decisions of keeping less quality.
In other words, (we can try, but) the results of encoding with -V 0 --resample 32Khz --lowpass 15 (-Y) will be different that those of -V 7. The encoder, at -V 0, will try to keep more precision of the remaining signal, and probably the bitrate won't be much higher.

So what j7n implied with his sentence is that just like we used to make fun of the " 128kbps is CD quality " (or when MS said it with wma and 64kbps),  it is not serious either to think "FM -> let's put the quality slider down, because there's no quality to keep".

Note: I don't imply that you said that.

[Edit: corrected quote]

FM Broadcast: 44 kHz or 48 kHz

Reply #19 –
I think I will do some tests with different settings, may be I need to see some audio diagrams to find exact lowpass number

As I understand real lowpass numbers is what I see in lame output ("18774 Hz - 19355 Hz"), and this range depends on --lowpass and  --lowpass-width, ie:
[ lowpass-%lowpasswidth .. lowpass+%lowpasswidth ]

so --lowpass is simply Arithmetic Mean from the target range (unuseful signal)

FM Broadcast: 44 kHz or 48 kHz

Reply #20 –
Quote from: ZinCh on
As I understand real lowpass numbers is what I see in lame output ("18774 Hz - 19355 Hz"), and this range depends on --lowpass and  --lowpass-width, ie:
[ lowpass-%lowpasswidth .. lowpass+%lowpasswidth ]

so --lowpass is simply Arithmetic Mean from the target range.

That's roughly correct. BTW, avoid the temptation to set the lowpasswidth too small. That can lead to nasty peaks and dips in the frequency response curve. Just stick with the default.

FM Broadcast: 44 kHz or 48 kHz

Reply #21 –
Quote
You'll need to set lowpass to 16 kHz. 19000 Hz is the center frequency of that signal, but there are also 2k wide sidebands of unknown purpose present.


Aren't those just guard bands to prevent triggering of the stereo decoder by uncorrelated signals close to the pilot frequency?

FM Broadcast: 44 kHz or 48 kHz

Reply #22 –
Quote
' date='Jul 28 2008, 18:23' post='579786']Note: I don't imply that you said that.
It's cool. I didn't think you were.

I do understand what you're saying, but by altering the lowpass frequency away from the standard for a given preset, surely you're making a net move away from optimised LAME encoding performance? This then infers that we know better than the developers and everybody who's ever taken part in a group listening test.

Looking at the -V settings as an efficiently optimised standard, isn't the one that happens to be tuned to fit the source material almost perfectly going to produce acceptable results to the majority?

Isn't a -V preset optimised to give the best possible performance in terms of quality vs bitrate for the bitrate used, ergo, acceptable results if you match the bandwidth of the preset to the source?

I'm used to the days when I'd always choose a tape speed on a reel-to-reel tape deck to suit the bandwidth of the source material in order to avoid wasting tape. 3 3/4IPS was always plenty for FM radio, but I wouldn't have dreamed of going below 7 1/2IPS to record a commercial audio CD or a good piece of vinyl. Isn't ignoring the -V settings effectively doing the opposite kind of thing to LAME, ie, taking control of the tape speed because we don't trust a preset that seems to fit?

Cheers, Slipstreem. 

PS Apologies for all of the questions. I'm just trying to get a firmer grasp on why -V settings would exist if they don't work the way I think they do.

FM Broadcast: 44 kHz or 48 kHz

Reply #23 –
Quote from: j7n on
You'll need to set lowpass to 16 kHz. 19000 Hz is the center frequency of that signal, but there are also 2k wide sidebands of unknown purpose present.

The 19 kHz pilot tone is a very pure sine wave and its frequency is controlled very tightly. There should be nothing else present between 17 and 21 kHz except a little bit of noise.

FM Broadcast: 44 kHz or 48 kHz

Reply #24 –
Maybe you're right. Then the sidebands what I and 2bDecided on one ocasion have seen are probably generated in the receiver. I just checked again and they indeed decrease in amplitude when switching to mono.

About the bitrate. Everybody knows that -V2 or -V3 is required for transparent results. Why would radio be fine at -V7? You say that we should trust LAME and make no own adjustments. Then let LAME decide how many bits are needed for the signal, and don't assume that the signal is "just radio".