just one thing bothering me ..

[HughBanton]

When you compute the slopes, make sure you get the ends of each waveform right.

Yes I thought of that too .. I need a 'wrap around' to the first sample from every last sample. My - you're sharp, Martin!

Interestingly I've so far ignored this issue in my current version; turns out the error is inaudible on every note exept no. 73. (Which being well beyond the keyboard may well be never needed anyway). Evidently the difference is normally very small, except on the top note where sample+1 is beyond the end of the array and so is being interpolated with 'garbage', so no surprise.

But I delight in 'tidying up' .. (OCD I expect ) so I'll definitely fix it all when I re-work it in my dll.

According to Wikipedia there is a <fmt-ck> format chunk in the file header

I've made a start with just that, but as you say no CPU loss simply copying Exonerate/Infuzion's Stereo Wave maker module, since it's green. I just fancied getting to the bottom of it!

I had the idea of hacking the header from the wav file that their module creates from my waves and pasting it in at the start, since it ought to be exactly the same as the one I need. (i.e. all the bytes that start with 'RIFF' and ends with 'data', inclusive). Couldn't make it work yesterday evening but I'll maybe persevere for a while.

Thanks again.

[HughBanton]

I succeeded in expanding my dll routine to generate an interleaved 'value/difference/value/difference/value ..' sequence for Martin's interpolator method, which uses a single stereo Wave Read - working well. And I've also adapted Exo's 'memSection' dll function (in the FlowstoneGuru dll tutorial) to make a superior stereo mem generator. These two elements together work way better than my original sample-interpolation arrangement using a pair of Wave Reads. Thanks for the tip Martin.

As regards Stereo Waves .. I might start a new topic on this. I tried pasting a genuine wav header ahead of genuine wave values, but Array to Mem doesn't take any notice of the header and just outputs the whole thing as per the entered array values. As far as I can tell there are just two main ways to generate a stereo wave in Flowstone (that is to say, one that Wave Read recognises as interleaved stereo and splits accordingly) : Load it from disk using WaveFile, or generate it with Mem Create, which has a stereo/mono bool input. Then it works.

Maybe Flowstone has its own stereo/mono format, but there do appear to be two distinct versions of a 'mem', one that is 'just an array', and the other that somehow contains the stereo/mono flag. That's my impression anyway, I can't tell what's going on.

I've gone way off my original (abstract!) topic, so that's about it here I guess!