|
|
|
|
|
|
| Section 5
Low Cost Speaker / Room Measurement |
|
| Section 5a Using a
PC (go to Section 5b Testing without a PC) |
|
| Downloads and Links | |
| Test Tones, all 44.1kHz, 16 bit .wav files | pink noise wav file for RTA
testing white noise wav file for use with spectrum analyzers mls test tone for impulse testing bong file for speaker testing CD1 1/3 octave modulated tones zip file. Use for testing without a pc. |
| Software Links, mostly free / open source W=Windows, M=Mac, L=Linux or BSD |
Audacity Open Source Sound Editor / recorder, (W,M,L).
It has a built in spectrum analyzer. http://audacity.sourceforge.net TrueRTA, excellent Real Time Analyzer (not free but only US$39 for 1/3 octave version) (W) http://www.trueaudio.com Praxis from Liberty Audio. Excellent, free RTA (W) http://www.libinst.com/praxis_downloads.htm |
| Low cost Test Microphones / sound meters | Behringer ECM8000, requires phantom power, see
microphone amplifiers below. Full audio frequency range. http://www.behringer.com/ECM8000/index.cfm?lang=eng Radioshack Analogue Sound Meter, Catalog #: 33-4050. Battery powered, self contained, response 50Hz - 10kHz, phono output. http://www.radioshack.com |
| Microphone Amplifiers. Required if your sound card does not have a phantom supply. | Behringer MIC100 http://www.behringer.com/MIC100/index.cfm?lang=eng ART Tube MP http://www.artproaudio.com/products.asp?id=1&cat=1&type=79 |
| Low cost but flexible Analogue Equalizers. May be used for basic room equalization. | Behringer T1951 (now discontinued) http://www.behringer.com/T1951/index.cfm?lang=eng TL Audio 5013 http://www.tlaudio.co.uk/docs_07/product_07/5013.shtml |
| Audio Testing Notes | |
| Audacity for beginners | 1.
Set sample rate and sample format in "Edit / Preferences" to 44100
and 16 bit 2. Create a new audio file with "File / New". Click record and the track should appear as a recording is made. 3. Stop the recording process after 3 to 4 seconds and drag the cursor over approximately 1 second of the recording. Select "Analyze / Plot Spectrum" 4. Here is a screenshot of the result so far. That's all we need to know at this point! |
| Measurement Methods | 1.
Test signals should be from one speaker only 2. I would suggest 3 speaker to microphone distances
3. When doing an actual recording of white or pink noise, aim for a recording level of -6 to -10db below peak. |
| Using the Spectrum Analyzer | Use Log Frequency scale and the Hamming Window. There is one parameter that needs to be changed and that is the sample size. Adjust this to the lowest value that gives the sharpest spectrum. For example, a 100ms selection at 44100 Hz requires a sample size of 8192. |
| Interpreting Results | 1.
Assume your speakers are flat in anechoic conditions and within the
manufacturers specification. Many listeners tend to blame their speakers for
any problems, not the room ! 2. Remember with a microphone distance >1.5m, what you are measuring, up to around 2kHz, is the the room response coloured by room effects. Room gain will increase bass output below 100Hz and room modes will give severe peaks and troughs in the same region. Floor reflections will probably give a major dip in the response in the lower midrange and crossover notches may also be present. 3. Mid Range and Tweeters Room effects in this range can best be explained with reference to the level of room absorption present. Remember, typical thicknesses of materials used means that absorption falls off rapidly below 500Hz and is low below 100Hz. To a varying degree, tweeters behave differently in rooms due to directivity effects. The more directional tweeters may maintain a flatter response but this will not match an omnidirectional midrange. Its response will suffer in a room with a fair amount of absorption as the reverberent field will be less and will collapse quicker. To summarise: Normal / bright room - poor bass, possible roll off above 2kHz Some extra room absorption - maybe more bass, tweeter response could go either way High room absorption (not in bass region) - poor bass, possible suck out in the mid range See this article for typical problem responses. |
| Speaker Impulse Testing | Impulse testing is particularly useful as it enables frequency
responses at higher frequencies to be measured in non anechoic conditions thus
eliminating the effects of the environment. We have included an mls test tone in the download section. This can be reproduced through your speakers and recorded in the same way as pink or white noise. The entire file has to be recorded with a small amount of silence at the start and finish. Unfortunately, deriving the impulse from the recorded tone requires specialised processing. The Audiophile Recordings Trust will process and return emailed files, for non commercial users, by special arrangement . A good quality speaker impulse and its spectrum is shown here. |
| Room Impulse Testing | A
room impulse can easily be measured by recording a balloon bursting. Hold the
balloon above your head while bursting in order to get omnidirectional results.
The Audiophile Recordings Trust will process and return emailed impulse files, for non commercial users, by special arrangement . This will give reverberation times over a range of frequencies. Here is a typical example recorded in a church hall with good acoustics for recording. |
| Making Improvements | Now
that we are able to make measurements, more options are available. Room treatment in the form of absorption, diffusers or both is easy to install at frequencies above a few hundred Hz and this should be considered first. Unfortunately, room effects in the bass region are difficult to treat in a domestic environment. However, room bass gain and other effects outlined in the "Interpreting Results" section, should be corrected by using an appropriately designed analogue equalizer. The nearest commercially available products I have found so far are mentioned in the links section above. They are for studio use and so are not ideal for use with typical unbalanced domestic equipment. Parametric equalizers like these examples will not introduce colouration if the filters are limited to a Q less than 1. |
| Section 5b Testing without a PC | |
The method uses a modified analogue sound meter, based on a Radio Shack product, and test tones that can be burnt to a cd or played back through a suitable sound card. However, if you have a pc (Microsoft Windows, Mac, Linux or BSD) a far more interesting and useful approach is to use the sound recording and processing abilities with that. Apart from the microphone, excellent free software is generally available to enable this. The modified sound meter is available in the UK from the Trust or Marshall Choong Audio (www.mcaudio.co.uk) The unmodified version can be purchased elsewhere but it is only specified from 32Hz to 10kHz. |
|
