Hacker News new | past | comments | ask | show | jobs | submit login
Ask HN: Why is Bluetooth audio always so bad?
58 points by pje on Aug 22, 2019 | hide | past | favorite | 75 comments
Pairing is famously a UX nightmare [0]. Even Apple's AirPods cut out constantly, suffer from interference from other nearby devices [1], and apparently just don't work in an open space [2].

To people with Bluetooth expertise: Why is the Bluetooth audio experience always so...bad? Are there some fundamental technical limitations in the spec? Are my expectations unreasonable? Or is it just always implemented poorly? Followup: are there competing technologies that Just Work™?

[0]: https://xkcd.com/2055/

[1]: https://support.apple.com/en-is/HT209369 ("move away from places [with] a lot of Wi-Fi activity")

[2]: http://www.iphonehacks.com/2018/04/heres-why-your-airpods-or-other-bluetooth-headphones-cut-out-while-crossing-a-street.html

So, no one has touched on a technical limitation. Bluetooth sucks for going through water. (If you like The Maths http://www1.lsbu.ac.uk/water/microwave_water.html )

Guess what people are mainly made of?

So, if you are standing between a BT transmitter and a BT receiver a lot of magic needs to happen to get the data moving between the two.

People love Airpods. People often put their phones in their pockets. A lot of water in the way.

If you notice BT being particularly terrible, look for how water might be blocking the signal.

This, of course, isn't the cause of all BT problems, but knowing this fundamental really helps troubleshoot when you run into a lot of them.

WiFi also uses 2.4 GHz (or even higher in newer versions) and it seems to be much more resilient. Is it really a matter of the underlying electromagnetic link?

Mostly bc WiFi has stronger signal, and the protocol is designed for higher speed and better stability.

I noticed this when in dense cities bluetooth is much worse when phone is in my back pocket. Moving it to front pocket makes a big difference.

> Are there some fundamental technical limitations in the spec?

It's the same issue as with practically everything else digital these days. Hugely complicated thousand-page standards that few, if any, understand in their entirety. Implementations that are rushed to market to "be fixed later because it's just software" - which never actually happens because the next thing comes along and instead of fixing existing shit it's again a rush to the deadline to push more of it out the door.

My dream is for society as a whole to decide we should just go back to the beginning and fix things starting from there until we catch back up with the present and blast right off into the future. For instance, agriculture. It really seems like there shouldn't be any problems. But even a farmer who's outstanding in their field will run into plenty of things that probably should have been solved centuries ago.

> outstanding in their field

I see what you did there.

As someone who has been working with IEEE 802.3 (a standard longer than 5000 pages) for the past four years I can tell you for every spec there are at least hundreds of people who understand the intricate details.

Whole hundreds, out of how many tens or hundreds of thousands of spec implementers creating crappy products?

It's a low power legacy frequency hopper with a bloated USB like network stack that isn't USB. That was the design from the very beginning.

It's unfixable.

My honest professional opinion is the people that wrote the spec had no idea what they were doing.

Is there anything better on the horizon? Is Bluetooth LE fundamentally related and thus also unfixable?

Would anyone at the time know what they were doing? Was there ever a similar protocol in existence yet? If the answer is yes, why did we need BT to be created?

20 years ago I was working on firmware for cheap narrow band spread spectrum cordless phone IC's.

It's easy to see the mistakes they made. If you look at the early to mid 1990's, there were cheap FM radio IC's that drew a couple of ma. At the same time ISM band microwave IC's were expensive, made of GaAs and just an LNA would cost you 50-100ma.

Wifi did exist back then. But the cost was around $150-200 for a chip set and the power requirements precluded battery operation.

So 'of course' they went with a low power FM radio.

The problem was as you shrink CMOS transistor sizes they get faster. By the late 1990's you could build a direct sequence transceiver IC out of CMOS. The one my group rolled out in 1999 drew about 200mA in receive mode and cost $4/1M

Zigbee came two three years later and the power requirements were down to 25-50mA.

Bluetooth designers completely failed to judge the progress being made in CMOS RF IC design. That was completely obvious to anyone in the industry.

The second problem had to do with the baseband. The Bluetooth baseband has several layers and essentially replicates the USB protocol/driver stack. Notice how easy it was for Microsoft to roll out the USB stack? Yeah total shitshow.

So you had a dumb simple frequency hopping radio married to a complicated baseband.

The rub, out of about two dozen companies that tried to build Bluetooth radios' circa 2000, all but three failed. Because they couldn't get the base band power consumption low enough.

Usually when you design a spec like Bluetooth you get your potential manufacturers on board working on silicon as you design the spec. They didn't.

Notable the Zigbee spec had the same problem with it's baseband and most people solved that problem by ignoring it and just using the radio.

A third problem (which effected Zigbee as well) is the transmit power was too low. Made worse with Bluetooth because the FM radio's sensitivity sucks. Back then I knew from cordless phone stuff that you needed at least 5db of transmit power to get reliable QOS. Bluetooth and first gen Zigbee radio's had less than 0db.

All of the above a small brained primate like myself could have told you.

> because the FM radio's sensitivity sucks.

It's worse then you think because Bluetooth is a frequency hopper which means Streaming QOS is limited by the link margin on your worst channel.

The core Bluetooth spec is 2985 pages. Essentially no-one correctly implements it; it's arguable that the spec is sufficiently complex, vague and self-contradictory that it's impossible to correctly implement. Unless you're using two devices by the same manufacturer (Earpods with an iPhone, a PS4 with the PS4 controller) your odds of getting a consistent, reliable experience are vanishingly slim.

I use airpods with an iphone and I constantly get connection issues, only one pod playing, or sound cutting off.

Is it possible you have a defective pair of AirPods?

yeah, they are my most reliable and longest distance BT device

If I remember correctly, the bluetooth headset profile (if you require microphone input) is pretty low bandwidth (think gsm quality), which would explain low quality in that case.

Newer Bluetooth variants support higher quality audio without a microphone (A2DP), and that's what you are likely using today unless you are in a call.

As far as interference in the 2.4GHz range, there is not much you can do about it in areas you don't control.

i'm basically agreeing:

> If I remember correctly, the bluetooth headset profile (if you require microphone input) is pretty low bandwidth (think gsm quality), which would explain low quality in that case.

the headset profile is a bit nicer than GSM (higher rates, more advanced codec), though nothing you'd want to listen to music through.

> Newer Bluetooth variants support higher quality audio without a microphone (A2DP), and that's what you are likely using today unless you are in a call.

A2DP has been around for awhile. as some other comments mention, there are multiple codecs that can be used. the minimum one is SBC (also used by HSP, at lower rates), which... isn't great.

once you're using aptX/AAC, perceived quality is far more likely to be dominated by the electrical and acoustic aspects of your headphones.

On that topic: why isn't there a way to connect headphones via A2DP and the microphone with HSP? It's rather disappointing that I have headphones with a built in mic, but I can't use the mic while using them as headphones.

Limitation of the particular Bluetooth version (4 vs 5?), needs to use one or the other. Believe it's addressed in a later version, either by a new profile or maybe similar to the solution you suggest. Did a little research into this after going nuts when headphone quality dropped when the mic turned on in a mobile game.

I'm still annoyed that I can't get FM radio to route over Bluetooth without getting some bizarre up.

My old Moto G could play FM Radio, and you could select Bluetooth as the output. Fast forward to today, and the Samsung A50 (in my hands right now), has the ability to pick the app, and select the output to Bluetooth, but it doesn't work for FM Radio. Why? Because Bluetooth is just not consistent in my opinion.

EDIT: I am probably in the minority of people that insist on FM Radio on their phone. Just putting that out there as most people will not run into my problems.

I have owned several mobile phones with FM Radio. With all of them the headphone cable provides the antenna.

Probably because FM is typically purely analog. You need to record signal, then compress it realtime and send through BT. That uses up battery. When using plain FM Radio, you just connect signal from fm chip to output amplifier.

That's a good point. The old Moto G did this out of the box. The newer ones, cannot. You need some "bluetooth router app". Beware of super spammy links if you google that. It's not an elegant solution and may or may not work depending on when Android gets updated.

Right now the A50 has a really nice sounding tuner with either my Samsung earbuds or Sony Headphones. NextRadio appears to be winding down it's features, so you have to use the basic tuner, and manually seek to each station, and add it to your favourites.

All this to say, there's not a lot of attention to FM Radio.

We could be using more advanced variants of DAB+/DVB/DMB protocols instead. It is still simple compared to WiFi but way more resilient than Bluetooth. But we get regulation in the way...

Don't phones typically use the headphone cable as the FM antenna though?

It isn't the case that your carrier might have deactivated FM, is it? If so, it might be possible to unlock this.


Yes, it is a shame that DAB is such a failure (at least in the UK it is, bit rate is like 80kbps or something). That plus bluetooth and it might be incomprehensible...

Bluetooth audio quality, while still noticeably worse than wired connections, is much better on iOS devices using the AAC codec than on earlier codecs. Check device specs before buying. Many currently available Bluetooth devices, even some very inexpensive ones, support AAC. Use aptX on Android, AAC on iOS. If you care more about quality than convenience, use a wired connection. If you are listening to contemporary popular music with extreme dynamic range compression (not the same as data compression), you might not even be able to tell the difference between a quality Bluetooth codec and a wired connection.

Recent versions of both iOS and Android include an improved SBC encoder that have a perceived quality equal to aptX, so, as long as you're using a modern device, the codec quality issue is largely moot.

In my Toyota, a wired connection from a phone's headphone out to the car stereo aux-in sounds terrible. The amplifier in the phone just can't send a loud enough signal to compete with the noise. Even a lossy Bluetooth connection from the phone to the car stereo sounds better in this case because it's not dependent on the phone's tiny amplifier to drive 6 speakers. The Bluetooth audio quality is almost indistinguishable from the high bit-rate MP3s stored on a USB drive.

Why is ogg opus not appropriate? Are there bluetooth accessories that support it?

I'm always very skeptical of audio quality claims. Did you happen to perform any blind test? They're very surprising.

Bluetooth is much better on Android/Desktops with support for apt-x HD low latency codecs.

Also LDAC on the Sony headsets

Yeah, LDAC is mostly supported in new versions of Android. It's really impressive on the Sony Wh-1000xm3 headsets.


Well Sony WH1000 MX2/3 perform really well, so do Bose QCII and the Harman Kardon Esquire also provide high quality sound.

BT headphones can and do sound nice.

My dad recently bought himself a WH-1000MX3, and after trying it myself, I wound up also buying one. Pairing it with my LG Android phone was no problem, and I'm quite impressed with both the audio quality (wired or Bluetooth) and the noise cancellation. They blow away my old MDR-NC60 set in both categories. Sony knocked it out of the park with this set.

It feels weird to shut them down and remove them in a room with modest background noise and feel, for a few seconds, that the noise is quite loud.

The only nitpick I have is, after reading the legalese, I decided that there was no way in hell I was going to use the Sony Connect app, but the phones sound so good to begin with that it's unnecessary to install it in any case.

I own a pair of Bose QCII and it cuts off when I'm walking down a busy street. Also, it maintains a connection to both my phone and laptop at the same time, but I can't hear the phone before I explicitly tell headphones to disconnect from the laptop via an app.

I do not know why it's so bad, but I do make a point to rant about it whenever I can (around friends). I've had terrible luck with it, absolutely never works for me. Another thing I've noticed about it which drives me crazy is that some devices will try to make up for issues by stretching/shrinking audio. This will either distort the tonality, or the rythm... but either way it messes up the song. And it blows my mind. Aux cords are cheap and foolproof.

In terms of alternatives, I have heard "zigby" mentioned. Also, aux cords Just Work. And I know this wouldn't work for everything, but I always wish I could have infrared transmission wherever possible. Even if I can't actually hear the transmissions, it still feels very loud to be blasting bluetooth out in every direction.

Do you mean ZigBee?


That's got to be the one. (https://en.wikipedia.org/wiki/Zigbee)

My friend who's an electrical engineer is quite fond of it. I really don't know more than his opinion

I haven't really heard of it being used for audio streaming, I think it's more suited for creating mesh networks meant for IoT purposes.

SBC codec, that is default and mandatory suck ass. It was chosen because they didn't want to pay the patent licensing fees of something better.

SBC was made for voice and is acceptable for that. Better codecs like AptX and AAC exists in many devices now, but because of incompabilites SBC is often negotiated anyway.

SBC can be quite good, but only if your devices can agree on a high enough bit rate. BlueZ on Linux can be prodded into negotiating a bit rate that's higher than default, and then it sounds pretty good, certainly good enough for BT speakers.

SBC is not a bad codec. It is a low complexity codec (sort of similar to MP2) optimized for low power consumption and quick encoding/decoding, and I think it's been unfairly maligned, due to devices that are way too stingy with bit rates.

ok, you are probably correct.

BT music from my iPhone to my car stereo sounds bad. There I cant verify what codec or bitrate is used, but clearly is a difference when comparing to a music file played from an SD card reader on the dashboard.

I also have a pair of Sony bluetooth headphones that I use with my MacBook, here the sound is awful, until I use a BT utility to choose aptX instead, that they also supports, but SBC is default selected anyway.

My BT headphones will generally use AptX when connected to my macbook, but when it tries to use audio from them, i.e. bluejeans randomly opens, it'll drop to SBC and immediately all my music sounds like junk.

Wish there's easy way to broadcast to multiple devices to share media. Had to use a bluetooth audio receiver with 3.5 jack + splitter + 2 cable headphones.

Also what is it about specific intersections that reliably causes connectivity / stutter issues? Is it something specific to the traffic signalling equipment or spectrum pollution? It happens more in dense urban environments but there are several extremely bland pedestrian crossings in my neighborhood where this happens as well. And then only on certain headphones.

Also the fact that battery reporting is still not standard is annoying.

I just bought a new car with BlueTooth integration, my first such car. I listen to Pandora when I drive.

I notice that the connection is not reliable, even when the phone is in a holder on the dash. I end up using the AUX input through the headphone jack.

I'm wondering if this is a common problem in cars, or if I should have my car looked at.

In my new car it works most of the time, but sometimes when I press the "Siri" button on my steering wheel, Siri will respond through the car's speakers. Then I'll ask her to play a song or navigate somewhere, and the audio starts playing out of my phone's speaker.

Fixing it requires going through the whole pairing and unpairing procedure. So same here, I just end up plugging it into the USB AUX.

Surveying audio support in cars over the last few years has taught me that car manufacturers have some of the worst tech around. It's almost universally bad.

Toyota seems to be less bad than most. I'm sure some others are also less-than-horrific. Maybe Tesla?

Interesting, mine is a Toyota 86 (which I guess is really a Subaru BRZ so I guess it's Subaru's Bluetooth stuff).

I drive a 2013 FRS (older US version of the 86) and I'll experience weird dropouts all day like once every couple of months. Otherwise it works flawlessly. Very strange. You might try using it now and again and see if the problem "just goes away". Very frustrating though. I did not opt for the touchscreen however. I think my head unit is made by Pioneer.

I have a Toyota. My second now. No issues with my iPhone and streaming.

As far as the UX nightmare, I'd attribute that aspect to the same problem EFI has: there are too many implementations, and most of them suck.

I think you've used only bad devices so you've had a bad experience. Use a good pair of Bluetooth headphones like Sony WH1000 or Bose Qc35 along with a good phone like Galaxy S10. It's perfect. No cut outs, amazing quality, insane range.

FYI, you might be shadow banned.

I know but I don't know what to do about it :(

Email hn@ycombinator.com and find out why they banned you.

it's really weird, because most of your (apparently perfectly ok) comments are dead, but some not, like this one.

It’s because I vouched for this one.

ok, but then you can't reply to dead messages, so you must have some kind of magic button!

That magic button is called “vouch”.

Go check the page count of the Bluetooth specs, and you'll have your answer.

Followup: are there competing technologies that Just Work™?

The W1/H1 chips, here after referred to as "Makes BT Not Suck" chips or MBNS, are the reason I put up with the otherwise unremarkable sound quality of Beats Studios and AirPods. I was reminded why just recently as I took the Bluetooth speaker out to the garage so the wife and I could do yoga. "Oh, yeaaaahhh. I have to go find the last device it was connected to, and disconnect it. Then, and only then, can I pair to her iPad."

If I had a HomePod or other AirPlay speaker in the garage, no matter what the speaker was connected to previous, I could have just told the iPad to use the speaker and be done with it.

It seems relatively easy to have Bluetooth not suck if you only have to build devices that don't suck with other devices you build. The original Wii Remote came out 13 years ago and worked great with the Wii over Bluetooth. Ditto for the 2007 DualShock 3 on the PS3.

I honest to goodness actually re-pair my Sony WH-1000XM3's every single time I use them. it's so lame.

Curious what you're pairing them with? Not had to re-pair my, er, pair at all. I have iPhone 6s and 2012 MacBook Air.

Sorry just saw this, using them with an iMac, iPad Pro, iPhone, and Macbook Air, it's not that they lose pairing but that they almost always connect to the wrong device.

How are airpod sound quality "unremarkable"? They sound good to me. . .

Not bad, not great. I've heard worse, I've heard better. They get the job done, but I otherwise have nothing of note to say about their sound quality. That's the thing about being unremarkable: I by definition don't have much else to say. :-)

ever listened to the shure se535?

Is there a BT device out there that you can't force into pairing mode?

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact