3 ms·

Buildings have a host of different floor arrangements. In elevators, you'll see P1, P2 (parking), L (lobby), B (basement), LB (lower basement), UB (upper baseme

Buildings have a host of different floor arrangements. In elevators, you'll see P1, P2 (parking), L (lobby), B (basement), LB (lower basement), UB (upper basement), M ("middle" floors, for buildings that have staggered floors), etc... Most elevators must know the name of the floor they're passing by (since they display it). But if you ship a speech system that just pronounces each symbol (e.g. "Floor pee one" or "Floor ell"), it may initially be confusing for most passengers ("What is 'floor ell'?"). A well-polished solution is going to take more effort by both the vendor and the client (and if it's not well-polished, the feature's likely to be considered a wart by those who don't actually need it). Plus, a beeper just needs to be audible - distortion is totally acceptable. A speech system will need a higher-spec'd amplifier and speakers (so that both components behave linearly over the full signal range) for the speech to be comprehensible. Far from insurmountable, but I think it'd be an error to say that there isn't more cost in announcing floor names than just beeping as you pass a floor. It's almost definitely marginal in the scheme of building costs, but it's there.

WalterBright 10y ago
> A speech system will need a higher-spec'd amplifier and speakers (so that both components behave linearly over the full signal range) for the speech to be comprehensible. No, it does not. I was surprised to find out that it was not necessary back in the PC era, where the speaker was simply connected to an I/O pin. It was meant to just produce buzzing noises. But programmers found out that it could produce surprisingly intelligible speech, at a modest but quite usable volume. How it's done is sampling a waveform of recorded speech at on/off levels at a rate at which data could be written to the I/O pin. The distortion inherent in driving a speaker on/off ironically smoothed it out into a decent approximation of the original waveform. Peoples' ears are extremely good at picking out speech from distortion, noise, etc. > it may initially be confusing for most passengers ("What is 'floor ell'?") No more confusing than a lighted "L" which I suggest remain. The audio is not a replacement for the lights, just a supplement. Though of course the audio could just as trivially say "Lobby" or "Parking Level 2" as "L" or "P2".

WalterBright 10y ago

> A speech system will need a higher-spec'd amplifier and speakers (so that both components behave linearly over the full signal range) for the speech to be comprehensible. No, it does not. I was surprised to find out that it was not necessary back in the PC era, where the speaker was simply connected to an I/O pin. It was meant to just produce buzzing noises. But programmers found out that it could produce surprisingly intelligible speech, at a modest but quite usable volume. How it's done is sampling a waveform of recorded speech at on/off levels at a rate at which data could be written to the I/O pin. The distortion inherent in driving a speaker on/off ironically smoothed it out into a decent approximation of the original waveform. Peoples' ears are extremely good at picking out speech from distortion, noise, etc. > it may initially be confusing for most passengers ("What is 'floor ell'?") No more confusing than a lighted "L" which I suggest remain. The audio is not a replacement for the lights, just a supplement. Though of course the audio could just as trivially say "Lobby" or "Parking Level 2" as "L" or "P2".