Comment by ansible

Yes. All the software / DSP processing can't help if the physics is against you.

By placing the microphone close to the mouth noises you make, you are increasing, by many orders of magnitude, the signal-to-noise ratio.

Cars and other large BT devices (like an Amazon Echo) can use multiple microphones and some fancy processing to really isolate the voice signal coming from a specific direction (relative to the device) and also cancel out surrounding noise sources.

With ear-bud systems, you can't fit in many microphones, nor can you spatially separate them enough for the software to do the job.