I am getting cracking noise with my multimedia project, I have hardware limitation that support only 16 bit integar value , where as incoming data is in float,
So i want to convert float to short with minimum precision losing i have used multiplication technique that is giving much lose, can any body help me for same?
Do you know the range of the incoming data values, i.e. minimum and maximum valid values? First approach would be uniform quantization of the floating point data values between minimum and maximum, with proper rounding. Beyond that, do you know the distribution of the data (e.g. measured approximately with a histogram)?
Not enough info for a definitive answer. So following may apply (or not).
If you are talking about "playing" the media on hardware that supports only 16 bit ints, then scaling, quantizing, etc., probably won't work UNLESS you can "tell" the hardware that the data has been scaled/quantized (and HOW it has been), OR if it already expects the data to be scaled/quantized (in which case, you'd have to scale or quantize the way that the hardware expects...)
In general, it would seem to me (again, not knowing the actual specifics involved in this case) that the best you'll be able to do when converting float to int is to "round" - as an integer has no fractional part (unless the hardware accepts scaling of some sort).
- Badges
- Science topic
- Similar topics
- Computer Science and Engineering
- Programming Languages
- Python