Timeline for Precision-related error in Audio
Current License: CC BY-SA 4.0
19 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Jun 19, 2018 at 19:22 | comment | added | Daniel W | @chuy and You were correct after all, "Real" did fix the problem, if it is typed in correctly, which I did not when testing it. The rounding problem, however, is an actual bug. I just missed getting an earlier solution by not typing carefully. | |
| Jun 19, 2018 at 19:18 | answer | added | Daniel W | timeline score: 2 | |
| Jun 19, 2018 at 19:08 | history | edited | Daniel W | CC BY-SA 4.0 |
added bug tag and header
|
| Jun 12, 2018 at 0:11 | comment | added | halirutan |
@DanielW After you get an answer from Wolfram acknowledging that this is a bug, please feel free to add the [bugs] tag. Additionally, it would be awesome if you edit your question then and include a "bug header" at the very top like in this question.
|
|
| Jun 10, 2018 at 2:57 | review | Close votes | |||
| Jun 12, 2018 at 0:11 | |||||
| Jun 7, 2018 at 23:20 | history | tweeted | twitter.com/StackMma/status/1004865677340106752 | ||
| Jun 7, 2018 at 18:41 | comment | added | MarcoB |
As @chuy mentioned, this may be OS and version-dependent. With your code above I get some 0.5, some 0.25. The problem goes away entirely if I use Audio[..., "Real"].
|
|
| Jun 7, 2018 at 17:37 | comment | added | Daniel W |
@MarcoB Integers. There may be others, but it seems to happen for all n in the Integers. First[AudioData[Audio[Sin[(25.*Pi)/6 + Range[-10, 10]*Pi]]]] gives a list of +-.0.25.
|
|
| Jun 7, 2018 at 16:40 | comment | added | MarcoB |
@DanielW for what values of n do you get the problem?
|
|
| Jun 7, 2018 at 14:53 | comment | added | Daniel W |
@MarcoB it only happens at specific frequencies for specific sample rates. There appears to be something special about Sin[(25.*Pi)/6 + n*Pi].
|
|
| Jun 7, 2018 at 14:30 | comment | added | chuy |
This might be OS dependent. In[104]:= AudioData[Audio[a], "Real32"] Out[104]= {{0.5, 0.5, 0.5, 0.5, 0.25, 0.25}} is what I get using 11.3 on Windows.
|
|
| Jun 7, 2018 at 14:14 | comment | added | MarcoB | @DanielW Thank you! v. 11.2 works fine (output), so the problem seems to have been introduced to 11.3. | |
| Jun 7, 2018 at 14:11 | comment | added | Daniel W | @MarcoB Good suggestion, done. | |
| Jun 7, 2018 at 14:10 | history | edited | Daniel W | CC BY-SA 4.0 |
Added code to generate jagged sine wave
|
| Jun 7, 2018 at 14:02 | comment | added | MarcoB |
I appreciate that you made a minimal example, but could you include the actual code that generated the sinewave with the jagged teeth as well? I can't reproduce it with simply First@AudioData@Audio@Table[Sin[x], {x, 0, 10, 0.1}] // ListLinePlot.
|
|
| Jun 7, 2018 at 14:01 | history | edited | MarcoB | CC BY-SA 4.0 |
Formatted code so it is easier to copy
|
| Jun 7, 2018 at 13:59 | comment | added | Daniel W | @chuy Thanks for the suggestion, but nope. I tried all available type settings, and the last two are always half the value of the first 4. | |
| Jun 6, 2018 at 20:47 | comment | added | chuy |
By default Audio uses 32 bit maybe try AudioData[Audio[a, "Real"]]?
|
|
| Jun 6, 2018 at 15:16 | history | asked | Daniel W | CC BY-SA 4.0 |
