Measurements of Bluetooth sound quality?

[Fluffy]

Bluetooth as a subject is very confusing, and from all the protocols and features it's hard to understand its actual effects on sound quality. Are there measurements (thd+n, sinad, FR, imd, etc.) of different Bluetooth applications that show what is the end result of whatever it's doing?

 

[wwenze]

Most standard measurement methods return a very good THD+N and SINAD because a single tone can be compressed with nearly no loss.

I've never tried 32-tone IMD tho, might be interesting tho I won't be surprised if that returns good results too.

 

[Webninja]

I haven’t checked in the last year, but my extensive research on Bluetooth (mostly for the car) resulted in the data rate being much less than CD quality. It might have improved with Bluetooth 5.

Bottom line, if you’re streaming Tidal or Qobuz (or CD/SACD, Flac) having Bluetooth in the music delivery chain will reduce quality.

 

[BDWoody]

I haven’t checked in the last year, but my extensive research on Bluetooth (mostly for the car) resulted in the data rate being much less than CD quality. It might have improved with Bluetooth 5.

Bottom line, if you’re streaming Tidal or Qobuz (or CD/SACD, Flac) having Bluetooth in the music delivery chain will reduce quality.

Question is what is audible.

 

[Webninja]

That’s true, but I did a quick check, and depending on codec used, still not even half of lossless.

You’re right that not everyone or every system will have it audible. I am ok with it in the car, but not for my headphones and more detailed listening.

 

[BDWoody]

That’s true, but I did a quick check, and depending on codec used, still not even half of lossless.

You’re right that not everyone or every system will have it audible. I am ok with it in the car, but not for my headphones and more detailed listening.

That's what I'm saying...half of lossless I'd bet is pretty close to testably transparent under 99% of conditions, but haven't done any proper tests myself.

Edit: I'm petty hopeless in determining differences at 320k... I was shocked after feeling like I MUST be able to tell...I'm a serious listener after all...

 

[Fluffy]

Please people, I didn't ask for opinions or estimations. I'm just interested in seeing actual measurements of the resulting signal after all the processing and transmitting done by a Bluetooth device.

If a single tone is not revealing enough because the nature of compression, then a multitoned test would be interesting to see, or even a Deltawave comparison to the original file.

 

[wwenze]

Spoiler

https://www.reddit.com/r/audiophile/comments/ciebxt

So, yup, it's sucking in IMD at least for one of them. The others can't really tell, might be the noise floor that is also increasing with frequency. Need the FFT plot to confirm. I remember the two-tone IMD to be pretty clean tho.

More digging:
https://www.soundguys.com/the-ultimate-guide-to-bluetooth-aptx-and-aptx-hd-19914/

SBC struggles with 2-tone, aptx is... will be hidden below the noise floor of most cheap receivers I guess. Neither showing any major harmonic distortion issue.

 

[mansr]

Spoiler

https://www.reddit.com/r/audiophile/comments/ciebxt

So, yup, it's sucking in IMD at least for one of them. The others can't really tell, might be the noise floor that is also increasing with frequency. Need the FFT plot to confirm. I remember the two-tone IMD to be pretty clean tho.

http://www.sereneaudio.com/blog/how-good-is-bluetooth-audio-at-its-best
Pure tones: (Note the noise floor)

Those graphs look FFTs with bad windowing. I wouldn't trust them.

 

[detlev24]

[...] It might have improved with Bluetooth 5. [...]

Unfortunately, it didn't with version 5.0. See:

"Bluetooth 5, Classic and Low Energy" on 'Audio over Bluetooth: most detailed information about profiles, codecs, and devices'.

[EDIT]
Changes from v5.0 to v5.1 to 5.2 (pages 333 and 336).

 

[mansr]

Please people, I didn't ask for opinions or estimations. I'm just interested in seeing actual measurements of the resulting signal after all the processing and transmitting done by a Bluetooth device.

If a single tone is not revealing enough because the nature of compression, then a multitoned test would be interesting to see, or even a Deltawave comparison to the original file.

Artificial test signals are mostly useless for evaluating perceptual codecs. Blind testing with carefully chosen music samples is the only sure way to test them. A perceptual model can produce an apparent distortion metric, but this will only be as good as the model. FFTs of tones will look fine unless the encoder has really screwed up. A multi-tone test isn't representative of real sounds and can make the encoder's perceptual model collapse, producing a poor output even if it performs well on music.

Good perceptual codecs are hard to create and harder to evaluate. That's why lossless is preferable whenever the bandwidth permits.

 

[wwenze]

(Wait, u can stuff MP3 into A2DP?)

Anyway... the problem is the codecs, and we haven't had any fundamental change or major revolution in lossless audio compression and I doubt we ever will. The benefits of BT5 are at the interface level. Like allowing you to stuff more bandwidth more reliably (so in practice, you may hear a difference if your sound is previously trash).

 

[ZolaIII]

LDAC:
https://www.soundguys.com/ldac-ultimate-bluetooth-guide-20026/amp/

 

[Webninja]

Comparison of SBC, MP3 and APTX :
https://www.sereneaudio.com/blog/how-good-is-bluetooth-audio-at-its-best

 

[Vincent Kars]

A good read: https://habr.com/en/post/456182/
Bit more by Vadikss: https://btcodecs.valdikss.org.ru/

 

[ZolaIII]

I think implementing Opus decoding & BT transfer would be a next logical step regarding BT audio regarding both latency and quality.

 

[Tks]

Gonna have to wait until bossman or wolfman get the BT module addon for the AP >_< Damn thing costs like $8K or something, yuck.

 

[bennetng]

Comparison of SBC, MP3 and APTX :
https://www.sereneaudio.com/blog/how-good-is-bluetooth-audio-at-its-best

Try the attached mp3, 96kbps only.