Experience with ISD or Quadravox -playback chips?
Scott M
Posts: 43
Basically, I want spoken voice and sound effects to be played under the control of a stamp. I know, Quadravox has just the thing, etc. I have one of those, and it works ok for my talking alarm clock, but it's not quite right for my current project.
The problem is, the Quardavox uses the cueing mode of the ISD chip. So if you tell it to play message 191, it scans over messages 0-190 to find 191. It scans quickly, but scanning still interjects a noticable delay, which is really, *really* annoying when you want it to speak complete·sentences. My alarm clock sounds like a drunk, trying to get a sentence out. Luckily, when I'm waking up, I'm not much more coherent myself, so I don't mind it so much. But my current project involves a lot of people who are wide awake and who aren't. going to. have. any patience. with a. circuit. that. can't keep a. sentence. going.
What I'd like to do is stuff my own messages into an ISD chip, and somehow know what addresses the messages are at, so I can tell the chip to play from that address, directly. No cueing, no funny pauses.
The problem is, I can't figure out how. You can stuff a message into any address you like, but there doesn't seem to be a way to tell where the last message you entered ends! So there's no easy way to know where to put your next message. I could do trial and error to find the ends of messages, but dear mercy that could take days.
Anyone have an elegant solution? I mean, I suppose I could buy some memory, a clock, a latchable counter for the address, and a DAC chip and figure out how to stream my own blasted audio from any address I like, but that doesn't sound like much fun. Advice appreciated.
The problem is, the Quardavox uses the cueing mode of the ISD chip. So if you tell it to play message 191, it scans over messages 0-190 to find 191. It scans quickly, but scanning still interjects a noticable delay, which is really, *really* annoying when you want it to speak complete·sentences. My alarm clock sounds like a drunk, trying to get a sentence out. Luckily, when I'm waking up, I'm not much more coherent myself, so I don't mind it so much. But my current project involves a lot of people who are wide awake and who aren't. going to. have. any patience. with a. circuit. that. can't keep a. sentence. going.
What I'd like to do is stuff my own messages into an ISD chip, and somehow know what addresses the messages are at, so I can tell the chip to play from that address, directly. No cueing, no funny pauses.
The problem is, I can't figure out how. You can stuff a message into any address you like, but there doesn't seem to be a way to tell where the last message you entered ends! So there's no easy way to know where to put your next message. I could do trial and error to find the ends of messages, but dear mercy that could take days.
Anyone have an elegant solution? I mean, I suppose I could buy some memory, a clock, a latchable counter for the address, and a DAC chip and figure out how to stream my own blasted audio from any address I like, but that doesn't sound like much fun. Advice appreciated.
Comments
http://www.parallax.com/dl/docs/cols/nv/vol2/col/nv65.pdf
http://www.parallax.com/dl/docs/cols/nv/vol2/col/nv66.pdf
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Jon Williams
Applications Engineer, Parallax
Dallas, TX· USA
It's a pity that ISD didn't spend the extra dollar to make the address lines read-write instead of write-only. That would have made everything really easy....
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Jon Williams
Applications Engineer, Parallax
Dallas, TX· USA
·· I may be wrong, but looking at the docs for my old ISD1000, I would guess that if you know how long the audio is that you're recording, and the start address, you should be able to calculate the end address, or be close enough where it won't take days to figure it out.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Chris Savage
Parallax Tech Support
csavage@parallax.com
I don't suppose Parallax could ask Quadravox to offer a non-cueing variant of their board, that lets you directly specify *and get back* the addresses for each sample? They've gone to the trouble of creating a serial interface; it could report this data back on demand...