Notifications
Clear all
Topic starter
A simple hi-hat sample I have gets this message when attempting to load it into the Yamaha MONTAGE.
This is the sample:
I’ve tried every modification I can to get it to load.
I lengenthed it out to 11.707 seconds to make sure there is no possible way it is too short.
Since that failed, I assumed it must be looking at the wave file itself and possibly trimming 0 values, so I added some noise at the end, plus a loop point with a few non-zero samples. I also converted from 44,100 Hz to 48,000 Hz.
Since that failed I know it must be doing more analysis on the sample than it should, so I normalized it to ensure it has the whole range of volumes.
This is shown in the image.
It’s loud, it’s long, it’s in both valid Hz, so why isn’t it loading?
To top off everything, this error isn’t even mentioned in the Yamaha MONTAGE owner’s manual.
It has the opposite error:
* Sample is too long. The Sample size is too large and the Load operation cannot be executed.
The Yamaha MOTIF XF owner’s manual has this error:
* Sample is too short. The Sample length is too short and the Frequency Convert Job cannot be executed.
But that gives me absolutely no clue how I am supposed to fix it, since I already made the sample insanely long, so it’s clearly not actually too short.
On the Yamaha MONTAGE side, I have rebooted several times and other samples still load after this one fails. There is plenty of sample memory available (around 200 megabytes) and no other points are suspicious. It is on firmware v2.00.
L. Spiro
Posted : 12/03/2018 7:12 pm
Not sure how Montage compares to Motif in this arena.
http://www.motifator.com/index.php/forum/viewthread/459440/
12 seconds (approx) is far below the Motif Max of 380 seconds.
Posted : 12/03/2018 7:50 pm
It’s loud, it’s long, it’s in both valid Hz, so why isn’t it loading?
Most likely the sample is too short, it finds that message when it cannot read the data... your adding sound to the end doesn't work after the sample reaches silence; there is a silence detection function. Adding some noise 10 and a half seconds later will not work to lengthen the sound. Silence detection is looking for silences of a couple hundred ms. Dead air shuts off playback.
It is likely the sample is too short.
Posted : 12/03/2018 10:58 pm
Topic starter
I believe we can confirm this is a bug.
Here is the sample:
https://www.dropbox.com/s/fdb0in61jl3mh3k/ffff.wav?dl=0
I downsampled it by 48 semitones to stretch the sound out.
Failure.
So I copied the loud parts of the sample over a lengthy period and then increased the loudness.
Failure here can lead to only 1 conclusion: There is a bug in the operating system.
(It failed.)
An image is attached.
The sample very clearly is long enough and loud enough. The selection alone is 3 seconds.
How can I proceed?
L. Spiro
Posted : 13/03/2018 12:35 am
Failure? What does that mean? (I think I know it means that you did not have success but it tells us nothing about what happened).
What does it report? Does it still say "Sample too short!"?
Posted : 14/03/2018 12:04 am
Topic starter
No update to the reported error was given because it remains the same.
No matter what I do I get, “Sample is too short.”
This was taken after attempting to load exactly the “ffff.wav” file linked above.
L. Spiro
Posted : 14/03/2018 2:43 am
Topic starter
44,100 Hz, 16 bits, 12.325 seconds long. The sample is linked above for any testing anyone wants to do on his or her own machine(s).
L. Spiro
Posted : 15/03/2018 12:41 am
Try stereo.
https://www.yamahasynth.com/forum/loading-wav-files-to-montage
Looking at the supplied WAV file, I see it is mono.
I'm not necessarily saying that mono is invalid - only that I've seen previous complaints which included trying mono WAV files and some documentation which seems to suggest stereo WAV files should be used.
Posted : 15/03/2018 1:41 am
Topic starter
I have around 300 samples that are all part of the same batch (all are 44,100 Hz, 16-bit PCM, mono, all created exactly the same way). Almost all of them work fine except just this one and maybe 2 more.
But since this error makes no sense I tried it anyway.
It still says, “Sample is too short.”
L. Spiro
Posted : 15/03/2018 3:37 am
Topic starter
I have reversed the sample, changed bit depth to 24, increased the volume to insane levels, made it stereo, cut, sliced, and diced it in numerous ways, etc. and I always always always get “Sample is too short.”
I am not sure what else I can try, and it is very clearly some kind of bug.
I have needed to record this song since Monday and I am completely blocked by this.
Can anyone else confirm this sample causes a bug in his or her own Yamaha MONTAGE that has it incorrectly reporting that the sample is too short?
How can I get past this to move forward on my project?
L. Spiro
Posted : 16/03/2018 2:30 am
Topic starter
I was able to load the original unmodified sample after saving it “without metadata”.
I don’t know what metadata was in the file that caused a bug, but all of my samples go through the same process and have the same metadata, and only a few cause this problem.
If Yamaha wants to look into this, the link above to a long and loud sample that triggers “Sample is too short” seemingly just because of some metadata is still valid.
L. Spiro
Posted : 16/03/2018 3:00 am
Topic starter
Before anyone wracks his or her brain too hard on it, I found that this and the other samples have start and end loop points at T=0 inside the file, which is why it had this error. It seems to sneak into non-looping samples after my tool chain first inserts valid loop point into a previous sample.
L. Spiro
Posted : 18/03/2018 4:41 am
Makes sense. Too short no matter what the data looks like. I didn't load with my tools that can look at loop points.
Wavosaur is the cleanest for me (from freeware tools listed here: https://yamahasynth.com/forum/are-there-any-applications-out-yet-to-create-you-re-own-waves-for-the-montage#reply-12856 ) to show loop points.
Information box shows loop is from sample 0 to sample 30. You can see "end loop" is near the start of the entire waveform.
This zooms into the start - also showing statistics for how many samples are in the full waveform (543,568).
Showing technically (due to amount of significant digits), sample 0 = Time 0:000 and sample 30 = Time 0:000. However, if more "time" digits were shown, you would see this is not the same "time". (30 samples/543568 samples)*100 = 0.006%. Meaning you are looping the first 0.006% of this full sample.
It (your issue) is figured out - but using this somewhat to remind me of what tools look at the loop points better than what I was trying to do before.
Posted : 18/03/2018 6:15 am
Topic starter
Wavosaur is also how I spotted this problem. You can see from my first screenshot that I added a loop point at the end, and Sony Sound Forge Pro 11 shows only that, hiding the secret NULL loop point at the start.
Wavosaur shows them as in your screenshots.
Now I’ve actually switched to using Wavosaur for almost everything I am doing.
I don’t know why it is so hard to find a good package for working with loop-point wave files.
L. Spiro
Posted : 18/03/2018 7:40 pm
Not sure why commercial software doesn't do a better job. The wav "chunk" format is well documented. There is some variability on interpretation, but your .wav file was well behaved. I wrote a quick python script that produced the same results as Wavosaur:
RIFF Chunk
File Size: 1087240
Rifftype: WAVE
Format Information Chunk
Chunk Sz: 16
Compression Code: 1
Number of chann.: 1 (mono)
Sample Rate ....: 44100
Avg Bytes/Sec ..: 88200
Block Align.....: 2
Sig Bits/Sample.: 16
... no additional bytes
Data Chunk
Sample Length: 1087136
Samples: 543568
Sample Information Chunk
Manufacturer...: 0
Product .......: 0
Sample Period..: 0
Midi Unity Note: 42 (F#1) / Yamaha Standard where note range is C-2 to G8
Pitch Fraction.: 0
SMPTE Format...: 0
SMPTE Offset...: 0
Num Smple Loops: 1
Sampler Data...: 0
Sample Loop Num = 1
Cue Point ID: 0
Type........: 0 = Loop Forward (Normal)
Start.......: 0
End.........: 30
Fraction....: 0
Play Count..: 0 (Infinite)
Posted : 19/03/2018 1:27 am
Page 1 / 2
Next
