Amplifier/Room measurements with audio recorder?

[RogersCompactMonitorFan]

I feel this might have been covered before, but I can't find an appropriate post...

TLDR: I'd like to measure CD player+amplifier+speaker frequency response and distortion using a Tascam SD-card audio recorder as I don't have a Microphone. Is this do-able relatively easily?

I've got a couple of amplifiers from the 1970s. One of them sounds "warm" compared to my Aiyima A07 amplifier, and I'm curious to find out why (or indeed, IF - is this my imagination? So I'd like to do Audio Science.)

Related: I have an old Sony CD player from the 1990s which I think sounds a bit distorted, and again I thought it would be good to measure to see if this is the case or if I'm imagining things.

While I'm thinking about measuring I thought it would be interesting to add speakers (and room) and do a comparative measure - this would need a microphone.

My thoughts are:

- Generate audio frequency sweep files and burn them to a CD to play on the CD player (I might already have some on an audio test CD I think I downloaded and burned once).
- Build a potential divider and dummy speaker load (I have 2x 3 ohm wire wound resistors which should make a 6 ohm load) and connect 2 additional potential divider resistors such that the amplifier output can feed into a sound card or audio recorder.
- Step 3 = Process the recorded audio files to derive frequency response/distortion measurements.

I have a Tascam DR-07 Mk2 audio recorder. It doesn't have a calibrated frequency response that I know of, but I thought it likely to be "flat enough" for speaker/room tweak comparisons. This would save spending the money on an UMIK1/similar microphone. The recorder can be battery powered, so should be OK with a differential speaker output from the Class D Aiyima amplifier.

I'm assuming a calibrated microphone would give "accurate" room results, but if I only want relative measures I assume I don't need the calibration so much.

The Tascam specs https://tascam.jp/int/product/dr-07mk2/spec claim:

24bit/96kHz recording
Freq response: 20Hz to 40kHz +1dB/-3dB(Fs 96kHz, EXT IN to LINE OUT, JEITA)
Distortion: 0.05% or lower(Fs 44.1k/4k8/96kHz, EXT IN to LINE OUT, JEITA)
SNR: 92dB or higher(Fs 44.1k/48k/96kHz, EXT IN to LINE OUT, JEITA)

...which "looks probably good enough" for what I want. I'm assuming if I believe I can hear distortion it will be much more than 0.05%, and I'm sure it's likely such a thing has a pretty flat response on the line input.

It looks as though REW has the ability to generate frequency sweeps, but I'm unsure if it can work with a WAV file from an audio recorder.

I can solder up some resistors, but I think the bit I'm missing is "software that runs on MacOS/Linux which takes a .WAV file and outputs statistics/charts"

It looks as though Audacity might have a distortion plugin - https://forum.audacityteam.org/t/distortion-analyzer/8218/21 - but I don't know if it works well in this application or not.

Has anyone else done something similar and can point to some tips?

Many thanks.

 

[DVDdoug]

- Build a potential divider and dummy speaker load (I have 2x 3 ohm wire wound resistors which should make a 6 ohm load) and connect 2 additional potential divider resistors such that the amplifier output can feed into a sound card or audio recorder.

Right. When measuring electronics you don't want to go-through speakers and a microphone.

I've got a couple of amplifiers from the 1970s. One of them sounds "warm" compared to my Aiyima A07 amplifier,...
...I'm assuming a calibrated microphone would give "accurate" room results, but if I only want relative measures I assume I don't need the calibration so much.

Of course if you bypass the speakers & mic you don't have to worry about that. Your record or soundcard will be "flat enough".

You don't need accuracy to measure frequency response differences... A 3dB difference is still a 3dB difference, etc. But you can't trust an uncalibrated mic for actual frequency response measurements.

If you are measuring room modes/resonances in the bass range, they tend to create gross variations so you may not need a calibrated mic to see the big bumps & dips. And again, you can trust the differences between two different speakers (or between two speaker positions).

 

[RogersCompactMonitorFan]

Thank you for your reply - I think this confirms what I was saying, which is nice.

There do seem to be packages such as https://rug.mnhn.fr/seewave/ which can analyse WAV files, but I'm wondering if there's something a bit more like REW/ more of a complete tool rather than requiring extra code out there that anyone knows of. Ideally, load a WAV file and get response plots and perhaps some kind of noise/THD figure out....

I have a turntable which would be interesting to analyse too. I assume at some point you could get calibrated records to play, but I'm assuming they are likely to be old and worn and less of use nowadays.

 

[kemmler3D]

I'm 80% sure REW can import sweeps from files so that should be fine. However, I'm less sure the built in mics on the recorder will be good enough for room correction. As a rough gut check, maybe do a close-mic recording of the woofer playing a sweep to see if the LF response is reasonably flat.

And building on @DVDdoug s comment, don't try to use it for any corrections above 300hz or so. Room correction relies on measuring absolute values, not just relative changes, so a non-flat mic makes the job quite hard.

Even "real" measurement mics can have misleading variations in frequency response. From that POV the UMIK-1 is quite a bargain given that you get an individual calibration file with it.

 

[RogersCompactMonitorFan]

Thank you for the advice. Good to know about REW - I'll download it and experiment.

I need to get some parts before making the amplifier cable so don't think I will get to the soldering immediately, but I will report back with what I find...