For downsampling (decimation) one normally uses low pass filter before downsampling (anti-aliasing filter). I have thought that FIR is the best choice here for the low pass filter implementation for this case. But I was reading this page: http://www.dspguru.com/dsp/faqs/multirate/decimation and it said in 2.3.1 under "How do I implement decimation?" "To implement the filtering part, you can use either FIR or IIR filters." Is this true? I remember reading or looking at a lecture where it said that FIR is the correct choice for this case, and IIR would be a bad choice, I do not remember now the exact reason now, been a while... Thought to ask the experts here what they think. thanks, --Nasser
using IIR or FIR to implement the lowpass filter for downsampling?
Started by ●March 5, 2012
Reply by ●March 5, 20122012-03-05
btw, the page does mentions, later on, the following "The fact that only the outputs which will be used have to be calculated explains why decimating filters are almost always implemented using FIR filters!" So may be then the first statment need to be rewritten. So, no problem. all is fine. --Nasser On 3/5/2012 4:34 AM, Nasser M. Abbasi wrote:> > For downsampling (decimation) one normally uses low pass filter > before downsampling (anti-aliasing filter). > > I have thought that FIR is the best choice here for the low pass > filter implementation for this case. But I was reading this page: > > http://www.dspguru.com/dsp/faqs/multirate/decimation > > and it said in 2.3.1 under "How do I implement decimation?" > > "To implement the filtering part, you can use either FIR or IIR filters." > > Is this true? I remember reading or looking at a lecture where it > said that FIR is the correct choice for this case, and IIR would > be a bad choice, I do not remember now the exact reason now, been > a while... > > Thought to ask the experts here what they think. > > thanks, > --Nasser
Reply by ●March 5, 20122012-03-05
On Mon, 05 Mar 2012 04:34:11 -0600, Nasser M. Abbasi wrote:> For downsampling (decimation) one normally uses low pass filter before > downsampling (anti-aliasing filter). > > I have thought that FIR is the best choice here for the low pass filter > implementation for this case. But I was reading this page: > > http://www.dspguru.com/dsp/faqs/multirate/decimation > > and it said in 2.3.1 under "How do I implement decimation?" > > "To implement the filtering part, you can use either FIR or IIR > filters." > > Is this true? I remember reading or looking at a lecture where it said > that FIR is the correct choice for this case, and IIR would be a bad > choice, I do not remember now the exact reason now, been a while... > > Thought to ask the experts here what they think.A FIR is nearly always the best choice. But there are no fundamental reasons why, for some particular task, that an IIR wouldn't be better than an FIR. But it would have to be a corner case. -- My liberal friends think I'm a conservative kook. My conservative friends think I'm a liberal kook. Why am I not happy that they have found common ground? Tim Wescott, Communications, Control, Circuits & Software http://www.wescottdesign.com
Reply by ●March 5, 20122012-03-05
On Mon, 05 Mar 2012 04:36:28 -0600, Nasser M. Abbasi wrote: (top posting fixed)> On 3/5/2012 4:34 AM, Nasser M. Abbasi wrote: >> >> For downsampling (decimation) one normally uses low pass filter before >> downsampling (anti-aliasing filter). >> >> I have thought that FIR is the best choice here for the low pass filter >> implementation for this case. But I was reading this page: >> >> http://www.dspguru.com/dsp/faqs/multirate/decimation >> >> and it said in 2.3.1 under "How do I implement decimation?" >> >> "To implement the filtering part, you can use either FIR or IIR >> filters." >> >> Is this true? I remember reading or looking at a lecture where it said >> that FIR is the correct choice for this case, and IIR would be a bad >> choice, I do not remember now the exact reason now, been a while... >> >> Thought to ask the experts here what they think.> btw, the page does mentions, later on, the following > > "The fact that only the outputs which will be used have to be calculated > explains why decimating filters are almost always implemented using FIR > filters!" > > So may be then the first statment need to be rewritten. > > So, no problem. all is fine.I disagree with the above statement. That may be the justification that the author uses, but the same general tradeoff exists between IIR and FIR filters when doing decimation as when doing any other task. One does get some computational improvement in an FIR because one can skip computations where one can't in an IIR -- but that may not necessarily outweigh the advantages of an IIR for specific cases. (Mostly, the FIR is going to beat the IIR hands down. But for cases where you're not decimating much, and you don't care at all about lots of unequal phase delay, or you don't want the "preecho" of an FIR filter -- using an IIR filter may be to your advantage). -- My liberal friends think I'm a conservative kook. My conservative friends think I'm a liberal kook. Why am I not happy that they have found common ground? Tim Wescott, Communications, Control, Circuits & Software http://www.wescottdesign.com
Reply by ●March 5, 20122012-03-05
On 3/5/12 11:35 AM, Tim Wescott wrote:> > One does get some computational improvement in an FIR because one can > skip computations where one can't in an IIR --kinda hard to do efficient polyphase filtering with IIR.> but that may not > necessarily outweigh the advantages of an IIR for specific cases. >if you're downsampling by a factor or 2 or 3, *maybe* the IIR would be more efficient. you would still compute 1 or 2 in-between samples that you would then just throw away. but the IIR might be so> (Mostly, the FIR is going to beat the IIR hands down. But for cases > where you're not decimating much, and you don't care at all about lots of > unequal phase delay, or you don't want the "preecho" of an FIR filter -- > using an IIR filter may be to your advantage). >you don't have to have pre-echo with an FIR. your FIR kernal might be a minimum-phase design rather than a linear-phase design. like the IIR filtering, this sample rate conversion would not be invariant with respect to direction. play the sound file backwards and downsample and you will not get the same set of samples (but reversed, of course) as you would from downsampling in the forward direction. that may or may not be important to you. -- r b-j rbj@audioimagination.com "Imagination is more important than knowledge."
Reply by ●March 5, 20122012-03-05
Tim Wescott <tim@seemywebsite.com> wrote:> On Mon, 05 Mar 2012 04:34:11 -0600, Nasser M. Abbasi wrote:(snip, quoting a paper)>> and it said in 2.3.1 under "How do I implement decimation?" >> >> "To implement the filtering part, you can use either FIR or IIR >> filters.">> Is this true?(snip)> A FIR is nearly always the best choice. But there are no > fundamental reasons why, for some particular task, that an > IIR wouldn't be better than an FIR.> But it would have to be a corner case.Note that the statement above doesn't say anything about better or worse. I would, then, say that the statement is true. If it said that you SHOULD use IIR, for example, then it might be questionable. Probably one should read the actual paper to see if other parts of the explanation might make a difference, but I don't see anything wrong as written above. -- glen
Reply by ●March 5, 20122012-03-05
On Mon, 05 Mar 2012 12:49:31 -0500, robert bristow-johnson wrote:> On 3/5/12 11:35 AM, Tim Wescott wrote: >> >> One does get some computational improvement in an FIR because one can >> skip computations where one can't in an IIR -- > > kinda hard to do efficient polyphase filtering with IIR.I'm not so sure. Upon what do you base this supposition? What is your chosen method for doing polyphase filtering in IIR, and what leads you to believe that it is the optimal way?>> but that may not >> necessarily outweigh the advantages of an IIR for specific cases. >> >> > if you're downsampling by a factor or 2 or 3, *maybe* the IIR would be > more efficient. you would still compute 1 or 2 in-between samples that > you would then just throw away. but the IIR might be so > >> (Mostly, the FIR is going to beat the IIR hands down. But for cases >> where you're not decimating much, and you don't care at all about lots >> of unequal phase delay, or you don't want the "preecho" of an FIR >> filter -- using an IIR filter may be to your advantage). >> >> > you don't have to have pre-echo with an FIR. your FIR kernal might be a > minimum-phase design rather than a linear-phase design. like the IIR > filtering, this sample rate conversion would not be invariant with > respect to direction. play the sound file backwards and downsample and > you will not get the same set of samples (but reversed, of course) as > you would from downsampling in the forward direction. that may or may > not be important to you.Well, true. -- My liberal friends think I'm a conservative kook. My conservative friends think I'm a liberal kook. Why am I not happy that they have found common ground? Tim Wescott, Communications, Control, Circuits & Software http://www.wescottdesign.com
Reply by ●March 5, 20122012-03-05
On Mon, 05 Mar 2012 20:50:11 +0000, glen herrmannsfeldt wrote:> Tim Wescott <tim@seemywebsite.com> wrote: >> On Mon, 05 Mar 2012 04:34:11 -0600, Nasser M. Abbasi wrote: > > (snip, quoting a paper) >>> and it said in 2.3.1 under "How do I implement decimation?" >>> >>> "To implement the filtering part, you can use either FIR or IIR >>> filters." > >>> Is this true? > > (snip) >> A FIR is nearly always the best choice. But there are no fundamental >> reasons why, for some particular task, that an IIR wouldn't be better >> than an FIR. > >> But it would have to be a corner case. > > Note that the statement above doesn't say anything about better or > worse. I would, then, say that the statement is true. > > If it said that you SHOULD use IIR, for example, then it might be > questionable. > > Probably one should read the actual paper to see if other parts of the > explanation might make a difference, but I don't see anything wrong as > written above.If it said you _should_ use IIR I'd be questioning it, for sure! -- My liberal friends think I'm a conservative kook. My conservative friends think I'm a liberal kook. Why am I not happy that they have found common ground? Tim Wescott, Communications, Control, Circuits & Software http://www.wescottdesign.com
Reply by ●March 6, 20122012-03-06
On Mon, 05 Mar 2012 12:49:31 -0500, robert bristow-johnson wrote:> On 3/5/12 11:35 AM, Tim Wescott wrote: >> >> One does get some computational improvement in an FIR because one can >> skip computations where one can't in an IIR -- > > kinda hard to do efficient polyphase filtering with IIR.I was kinda snippy the last time, but: I'm not so sure that it's all that bad. If you use a filter form that realizes all poles in the states and only applies zeros on the output side, then I think you could do polyphase filtering with IIR, and do it fairly efficiently. I'm pretty sure that this would mean that you'd have to implement the filter stages in parallel, rather than in cascade; this, in turn, would mean that the coefficients for going from the states to the output would have obscure (and possibly lengthly) derivations. But I think it could be done. But then -- the OP didn't mention polyphase filtering, just decimation. If you're decimating by an integer amount, then you're not going to be doing polyphase filtering at all -- just ignoring some of the output. -- Tim Wescott Control system and signal processing consulting www.wescottdesign.com
Reply by ●March 6, 20122012-03-06
On 3/6/12 12:23 AM, Tim Wescott wrote:> On Mon, 05 Mar 2012 12:49:31 -0500, robert bristow-johnson wrote: > >> On 3/5/12 11:35 AM, Tim Wescott wrote: >>> >>> One does get some computational improvement in an FIR because one can >>> skip computations where one can't in an IIR -- >> >> kinda hard to do efficient polyphase filtering with IIR. > > I was kinda snippy the last time,don't sweat it. i am on and off here spottily. and i don't think you're one of the snippy's here at comp.dsp. whether or not i'm in that group is for other people to judge.> but: I'm not so sure that it's all that > bad. If you use a filter form that realizes all poles in the states and > only applies zeros on the output side, then I think you could do polyphase > filtering with IIR, and do it fairly efficiently.well, you don't have to compute the feedforward paths for output samples you don't keep, but you do have to do it for the feedback paths for *every* intermediate sample. the issue is: what if the upsample ratio is very large? like say, what if you're converting from 48000 to 44100? you have to upsample by 137 (stuffing in 136 zeros), LPF, and then throw away 159 out of 160 samples. with FIR you just drop anchor at, pick the correct set of coefficients (that's the "polyphase" thingie), and do a, say, 32-tap FIR. with IIR you have to do *something* for each of those 160 samples, including the 159 you don't keep. but if the OP is simply downsampling by a factor of 2 or 3, i think that an IIR LPF (and tossing out 1 or 2 samples) might be more efficient than a 32-tap FIR LPF.> I'm pretty sure that this would mean that you'd have to implement the > filter stages in parallel, rather than in cascade;dunno what you mean here, Tim.> this, in turn, would > mean that the coefficients for going from the states to the output would > have obscure (and possibly lengthly) derivations. But I think it could > be done. > > But then -- the OP didn't mention polyphase filtering, just decimation.right. which is the *latter* portion of the chore of polyphase filtering. essentially what the OP is doing is polyphase filtering without the step of upsampling and zero-stuffing. everything else is the same: the LPFing and picking out the samples you wanna keep.> If you're decimating by an integer amount, then you're not going to be > doing polyphase filtering at all -- just ignoring some of the output.well, you're doing *part* of what polyphase filtering is about. conceptually, with polyphase filtering, you are upsampling, then LPFing, then ignoring some of the output. doing it with an FIR saves you from unnecessary computations of output samples that you know a priori are ignored. but the IIR has still has to go through those ignored samples because it's needed to update its states. -- r b-j rbj@audioimagination.com "Imagination is more important than knowledge."
