I can imagine adda converters are improving fast.
I know you did not ask my opinion. But I will give it anyway.
For a lot of system components (AD/DA, Op-Amp's, line drivers, headphone drivers, Amplifiers, even passives), the low end, low cost components are dramatically improved over what was available in, say, 1990.
There is very little complete garbage left. Also, manufacturers got smart and first of all limit device features and flexibility and issue very limited datasheets and avoid complex and "interesting" Reference designs.
This means every engineer (even the Chinese Kopy Kat) simply copies the data-sheet/reference and reliably gets the datasheet performance. So in many ways the bar at the bottom of performance has been raised substantially.
For example, with OPA1678/79 available to pretty reliably replace NE5532, TL072 & TL074 in "commodity" audio designs at very low cost, there is no excuse to use these rather poorly performing dino's.
At the same time anything that we could call genuinely "High Performance", that is well above average, has basically disappeared from parts catalogues, be it IC's or passive parts, or is priced well past "la-la land" territory and usually not meant for Audio (e.g. 20 Bit / 1MspS Multibit ADC's and DAC's) making them hard to use.
As a result audio quality has kind of settled generally at a level that is high compared to low end products from over 30 Years ago, but at the same time is WELL, WELL, WELL below the best designs available 30 Years ago. And it has become hard to even make anything that approaches the subjective audio quality of the best designs of yesteryear.
Further, to illustrate Sturgeon's Revelation and Sod's Law the current day "copy & pasta" "engineers" manage to find whole new levels of messing up designs by not thin king things through.
As an example (also at the bottom range of products) let's take a common Bluetooth or Wifi Speaker. The BT and/or WiFi SOC nowadays are often in principle capable of 24 Bit / 48 kHz (aptX-HD) and 24 Bit / 96kHz (LDAC) Audio, while WiFI usually at least manages 24 Bit / 192kHz.
The Bluetooth/WiFi SOC/Module has a 90dB dynamic range, very poor audio quality AD/DA on board and ~2nS RMS Jitter which will degrade a 16 Bit dynamic range from 96dB to 70dB (we are nearly back at K7 levels) in "high quality" mode that needs to be explicitly enabled (and up to 20nS RMS Jitter in the default "power saving" mode).
Now both I2S and SPDIF outputs are possible but need modified firmware and modifying BT Chip firmware is slightly non-trivial. And 192kHz DSP hardware is expensive as are ASRC.
So what I see in many BT/Wireless Speakers I tore down is that the analogue output from BT and/or WiFI modules (usually with similarly bad AD/DA and high jitter) is then applied to the AD build into a 48kHz max DSP Chip, which is often of a similar quality as what is other SOC's.
The Output commonly passes through the DA on the DSP Chip and is then applied to a low end, analogue input Class D Amplifier Chip, often with excessive gain on the amplifier commonly followed by passive crossover speakers (single cap on tweeter).
The DSP programming often tries to do too much with too little DSP horsepower, leading to compromises in DSP operation. Or the programmer is simply doing "textbook" because he is just a coder who fails to understand audio, music etc.
So in principle, each device in the chain can deliver acceptable performance if hardly high quality used stand alone.
But daisy chained brainlessly the performance is lowered sufficiently, that any upper end 1980's boombox playing cassette tape with DolbyHx and Metal tape will be objectively and subjectively ahead!
Using literally mostly the same parts and switching the amplifier IC's to digital input, we can make a fully digital chain instead, but I see it very, very, very rarely in teardowns or service manuals.
By adding a low cost ASCR/Jitter-Cleaner we can ensure we accept all input signals and output a digital audio signal at 24/48 or 24/96 depending on DSP and Amplifier IC's used (fixed rate in DSP simplifies the DSP design).
Even low cost ASRC's nowadays have dealt with most issues and are sonically quite transparent and output 50ps or less Jitter (17 Bits or better DNR) using internal PLL, even less using Crystals for the output. Cost can be as low 2 USD for a 32 Bit / 384 kHz capable part in QFN32.
Naturally, crossovers are best realised digital in this case, which also affords us time/impulse correction. And lets spend an extra dollar or two (that we save elsewhere) by using a better, digital I/O only DSP Chip with professional mastering grade DSP Routines.
Now, using 80% identical parts, likely have a neutral budget, we have shifted from a system with very low actual dynamic range, to one limited by the Power Amplifier IC's.
The digital input power amplifier IC's I would use, manage 110dB(A) or better DNR @ ~ 20W/6Ohm @ 1% THD power output (and 0.01% THD&N from 0.1W to ~ 5W, H3 dominates HD) and can be used without or with minimal output filter while passing EMC easily.
And we have created a transient perfect system and overall improved acoustic system, without actually touching the acoustic system (except adding a little damping material, which is often considered optional).
Which system will sound better? Pretty obvious really.
While semi-pro recording/playback devices CAN be better, many are designed in the same way as the BT Speakers I mention, at the bottom end of the range, and perform accordingly.
Anyway, my two Thai Baht on the subject. Try not to spend all in one place, which is kinda difficult seeing a bottle or beer is 80 Baht at a cheap "Night Market".
Thor
