Dear Junko, I am not sure what you really wanted to say since you are describing two roads that are fairly well known and established:
1 Phones (or any other product) requiring a high-quality (Hi-Fi) audio uses a separate audio processing chip (DSP-based codec/coder-decoder).
Lower audio quality end-products integrate this function together with Power management (Dialog and Wolfson’s specialty, and Maxim’s newer business)
2 Two distinctly different “stories” (roads): (1) focus on a broad system-level integration (“jack-of-all-trades” in the chip) versus (2) focus on analog (audio) performance. Both require lowest possible power for mobile devices.
3 Cirrus (and Wolfson) has “married” the two “stories”: Integration of all audio-relevant (only) functionality as well as the commonly required minimal power. Namely, Audio subsystem integration (Audio SoC), without compromising the top-level analog audio performance. This need is driven by an increasing functional complexity and audio performance requirements of mobile devices (“theater-in-a- pocket”, so to speak).
But, as usual, great coverage and topics
Listening to heavily compressed audio through medium quality earbuds ... I don't think most consumer care about audio quality. My home audio set uses 10+ year old components (Marantz and B&W803) that are still ages ahead of iPhone quality. iPad speakers are terribly bad. I also wonder why people do not spend on better earbuds, already $40 will give a huge improvement.
"I sat in a recording session with three of them once and they were arguing and fiddling with the EQ for a couple of hours before they arrived at a sound they thought was acceptable. I never could tell the difference."
I can believe it. There's a story about the late Les Paul, guitarist and guitar designer, visiting a recording studio where his audio engineer son was recording a session. He listened for a moment and said "You're a fifth flat." His son was disbelieving, but instrument measurement proved dad was right. Les had perfect pitch, and heard a variance most folks would not.
Age is also a factor: the ability to hear high pitched sounds deteriorates with age. Those younger audio engineers probably did hear things you couldn't.
What you play back on is also a factor. The late Frank Zappa used to do his final mix-down using KLH 5 speakers instead of the usual JBL studio monitors, because he figured that was representative of what his audience would be using to listen to his albums, and he wanted it to sound good to them.
Good point, Frank. We will know more, when Samsugn actually starts u sing Dialog's PMIC with integrated audio. (The annoncement was made only recently) My understanding is that Samsung's currnet Galaxy S3 is using Wolfson's audio chip with Maxim's power management chip.
"...when you reduce the power on the audio chip, you degrade the sound quality."
Generally true, but I think the real difficulty of integrating system power management and audio into a single chip is that system power management is a big noise-maker -- switching regulators running at high currents.
Putting those things on the same substrate as an audio CODEC and maintaining the same audio performance as a stand-alone CODEC is a real challenge.
All make sense. But Apple has a track record in the past using both Wolfson's chip and Cirrus' chip in their different iterations of iPod.
I am not certain how the changes happened; but I do see a lot of iPod users are commenting that one audio codec was better than the other...
The difficulty in integrating power management and audio into a single chip has been around for quite a while. No one has really solved the problem because when you reduce the power on the audio chip, you degrade the sound quality. Some companies, like Wolfson, have in-house sound studios where audio engineers listen what the chips produce and then decide which is best. I would imagine that customers, like Apple, have their own team to determine sound quality. It seems to be a highly subjective process for the untrained ear, but having raised a budding sound engineer they can hear things that most of us miss. I sat in a recording session with three of them once and they were arguing and fiddling with the EQ for a couple of hours before they arrived at a sound they thought was acceptable. I never could tell the difference.
And in my experience with the audio chip industry, I've learned that each audio chip has its very own sound signature. Changing the chip means changing the end product. That, in itself, makes switching out Cirrus nearly impossible. And the effect of power management on the sound the chip produces would have to be absolutely nil to pass muster for Apple.