As has been described, a ‘perfect’ step response indicates a transducer that behaves perfectly in the frequency and time domains i.e. it duplicates the signal in terms of sound pressure; it is neutral. But with this simple statement comes a number of qualifiers:
But as I ranted on about earlier, I think the significance of the step can be appreciated logically as long as the designer knows what they want to achieve. In my case it is: a neutral, moderately directional speaker in a real room; if the combined output of the drive units stays reasonably time-aligned off axis and at different distances then that will be a bonus (another one of your tests as I recall, Ray). There’s not much point in my measuring this, though: my speakers already have the drive units in a vertical line as close together as possible on a baffle. Their geometry defines what they are going to do off-axis in a predictable way, and I can’t change that very much with DSP.
However, I am reasonably sure that this configuration is the best compromise for me in terms of the interaction with the room: it is moderately directional thus keeping reflections down to some extent; but it is small so I feel it will sound reasonably ‘natural’ in a real room. As such I won’t be investigating panel speakers, or phased arrays (unless it’s German-style bass cardioid), or truly omni-directional speakers, and any DSP room compensation I apply will be minimal. For me, I think I do want a real step i.e. time-aligned output from the direct sound at the listening position and I am happy that I will hear past any reflections to that, even if the in-room (non-anechoic) measurements look messy because of the reflections.
- the step may only be valid at a single point in space. If so, is this any great achievement?
- most audio people don’t believe that the time domain matters anyway.
- most DSP people don’t believe that a neutral speaker sounds good in a real room – if ‘room correction’ is applied, the step will be modified in some way.
- a good step derived from a sine sweep may not show how the speaker behaves with a real step (although Ray has demonstrated a good match with one of his tests), or at different volume levels
- the step cannot be expected to remain uncontaminated once room reflections reach the listening/measurement position. If step-iness is valued literally at the listener’s ears, reflections must be eliminated as much as possible:
- acoustic treatments = anechoic chamber
- use of headphones
- sit close to the speakers
- use a phased array to produce a directional beam of sound to the listener
- a desperate measure would be to kill the reflections at the listening position using DSP from the speaker but this would be very unstable, not possible in some rooms, and only valid at a single point in space.
But as I ranted on about earlier, I think the significance of the step can be appreciated logically as long as the designer knows what they want to achieve. In my case it is: a neutral, moderately directional speaker in a real room; if the combined output of the drive units stays reasonably time-aligned off axis and at different distances then that will be a bonus (another one of your tests as I recall, Ray). There’s not much point in my measuring this, though: my speakers already have the drive units in a vertical line as close together as possible on a baffle. Their geometry defines what they are going to do off-axis in a predictable way, and I can’t change that very much with DSP.
However, I am reasonably sure that this configuration is the best compromise for me in terms of the interaction with the room: it is moderately directional thus keeping reflections down to some extent; but it is small so I feel it will sound reasonably ‘natural’ in a real room. As such I won’t be investigating panel speakers, or phased arrays (unless it’s German-style bass cardioid), or truly omni-directional speakers, and any DSP room compensation I apply will be minimal. For me, I think I do want a real step i.e. time-aligned output from the direct sound at the listening position and I am happy that I will hear past any reflections to that, even if the in-room (non-anechoic) measurements look messy because of the reflections.