Is this Compatible with Raspberry Pi 3 Model B or only the standard Pi3?

Failing at the first hurdle in creating audio with:

2018-05-30 10:23:08.563832: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1053] Created TensorFlow device (/job:localhost/replica:0/task:0/device:GPU:0 with 10413 MB memory) -> physical GPU (device: 0, name: GeForce GTX 1080 Ti, pci bus id: 0000:06:00.0, compute capability: 6.1)
INFO:tensorflow:Unexpected error happened: Negative dimension size caused by subtracting 1 from 0 for 'Conv2D_2' (op: 'Conv2D') with input shapes: [1,1,0,128], [1,1,128,128]..

I have 2 instruments assigned per corner and all the WAV files look good according to soxi output.

Love this project, thanks for sharing!

As much as I admire having a physical device I was wondering, how hard it would be to make this work in software only. Would be nice to play with this without having to go through the process of actually building the hardware.

Would that be hard to achieve?
Has anyone looked into that possibility?

Hi guys, I already bought a Pi 2 because I found it cheap on craigslist, but I later realised this project calls for a RP 3. Is there anyway I can make it work or did I just throw money out the window? Cheers!

Hi. I would love to have some recommendations on websites where to produce the PCB from the files provided.

Set up the rpi 3 b+

• get latest raspbian as described on https://www.raspberrypi.org/downloads/raspbian/

• Do the things in section 3 and 4 from https://github.com/googlecreativelab/open-nsynth-super/tree/master/linux#3-prepare-the-os

get nsyth from git

$ wget https://codeload.github.com/googlecreativelab/open-nsynth-super/zip/master
$ unzip master && rm master

Set up the firmeware

• follow the guide from here https://github.com/googlecreativelab/open-nsynth-super/tree/master/firmware#installation-guide -> get en.stm32cubef0.zip from http://www.st.com/en/embedded-software/stm32cubef0.html and put in /home/pi/open-nsynth-super-master/firmware/

$ cd /home/pi/open-nsynth-super-master/firmware/
$ unzip en.stm32cubef0.zip 

• get the depencies
  $ cd /home/pi/open-nsynth-super-master/firmware/utils && sudo ./install_dependencies.sh
  -> this takes a while

• don't do this $ mv STM32Cube_FW_F0_V1.9.0 cube it gives then the error
  make: *** No rule to make target '../STM32Cube_FW_F0_*/Drivers/CMSIS/Device/ST/STM32F0xx/Source/Templates/gcc/startup_stm32f030xc.s', needed by '../bin/main.elf'. Stop.

• flash chip
  $ make install

sudo /home/pi/open-nsynth-super/firmware/openocd/bin/openocd --file openocd.cfg --command "program ../bin/main.elf verify reset exit"
Open On-Chip Debugger 0.10.0 (2018-05-14-21:26)
Licensed under GNU GPL v2
For bug reports, read
	http://openocd.org/doc/doxygen/bugs.html
BCM2835 GPIO nums: swclk = 25, swdio = 24
BCM2835 GPIO config: srst = 23
srst_only separate srst_gates_jtag srst_push_pull connect_deassert_srst
adapter speed: 1000 kHz
adapter_nsrst_delay: 100
srst_only separate srst_nogate srst_push_pull connect_deassert_srst
cortex_m reset_config sysresetreq
srst_only separate srst_nogate srst_push_pull connect_deassert_srst
adapter_nsrst_delay: 100
adapter_nsrst_assert_width: 100
Info : BCM2835 GPIO JTAG/SWD bitbang driver
Info : SWD only mode enabled (specify tck, tms, tdi and tdo gpios to add JTAG mode)
Info : clock speed 1001 kHz
Info : SWD DPIDR 0x0bb11477
Info : stm32f0x.cpu: hardware has 4 breakpoints, 2 watchpoints
Error: stm32f0x.cpu -- clearing lockup after double fault
Polling target stm32f0x.cpu failed, trying to reexamine
Info : stm32f0x.cpu: hardware has 4 breakpoints, 2 watchpoints
    TargetName         Type       Endian TapName            State       
--  ------------------ ---------- ------ ------------------ ------------
 0* stm32f0x.cpu       cortex_m   little stm32f0x.cpu       halted
adapter speed: 1001 kHz
target halted due to debug-request, current mode: Thread 
xPSR: 0xc1000000 pc: 0xfffffffe msp: 0xfffffffc
adapter speed: 1001 kHz
target halted due to debug-request, current mode: Thread 
xPSR: 0xc1000000 pc: 0xfffffffe msp: 0xfffffffc
adapter speed: 4061 kHz
** Programming Started **
auto erase enabled
Info : device id = 0x10006444
Info : flash size = 32kbytes
wrote 12288 bytes from file ../bin/main.elf in 0.614188s (19.538 KiB/s)
** Programming Finished **
** Verify Started **
verified 11788 bytes in 0.050718s (226.975 KiB/s)
** Verified OK **
** Resetting Target **
adapter speed: 1001 kHz
shutdown command invoked

copy audio files

• https://github.com/googlecreativelab/open-nsynth-super/tree/master/linux#4-copy-audio-files

Set up latest openframework from git

• get latest openframeworks from git
  $ cd ~ && mkdir opt/ && cd opt/ && git clone --depth=1 https://github.com/openframeworks/openFrameworks.git
• rename the openframeworks folder to be compatible with manual
  $ mv openFrameworks of
• get openframeworks dependencies
  $ cd of/scripts/linux/debian && sudo ./install_dependencies.sh
• get libs for openframeworks
  $ cd ~/opt/of/scripts/linux && download_libs.sh
• copy the files from https://github.com/googlecreativelab/open-nsynth-super/tree/master/app/open-nsynth to $ mkdir /home/pi/opt/of/apps/open-nsynth && cp -r /home/pi/open-nsynth-super-master/app/open-nsynth /home/pi/opt/of/apps/open-nsynth/
• replace in the code Poco::FastMutex with std::mutex -> this is in various files
• add ofxPoco to addons.make
• build the app in ~/opt/of/apps/open-nsynth/open-nsynth with
  $ make -j4 -> this takes a while
• get the settings.json in the right place https://github.com/googlecreativelab/open-nsynth-super/tree/master/linux#4-copy-audio-files
• test if it's running with
  $ bin/open-nsynth

get nsynth dependencies

$ cd ~/open-nsynth-super-master/linux

• edit ./nsynth-setup.sh and edit the section replace every gstreamer0.10 with gstreamer1.0

install_deps() {
    apt-get update
    apt-get install -y i2c-tools python-smbus gdb-arm-none-eabi gcc-arm-none-eabi \
        git autoconf libtool make pkg-config build-essential \
        libcairo-dev gstreamer1.0-dev gstreamer1.0-x \
        gstreamer1.0-plugins-base-apps gstreamer1.0-alsa \
        libudev-dev libsndfile-dev libopenal-dev libssl-dev \
        gstreamer1.0-plugins-good gstreamer1.0-plugins-bad \
        gstreamer-plugins-base0.10-dev freeglut3-dev libasound2-dev \
        libxmu-dev libxxf86vm-dev libgl1-mesa-dev libglu1-mesa-dev \
        libraw1394-dev libudev-dev libdrm-dev libglew-dev libopenal-dev \
        libsndfile-dev libfreeimage-dev libcairo2-dev libfreetype6-dev \
        libssl-dev libpulse-dev libusb-1.0-0-dev libopencv-dev \
        libegl1-mesa-dev libgles1-mesa-dev libgles2-mesa-dev libassimp-dev \
        librtaudio-dev libboost-filesystem-dev
}

• remove the follow section:

setup_openframeworks() {
    if ! [ -e /home/pi/opt/of ]
    then
        echo "Fetching openFrameworks"
        mkdir -p /home/pi/opt
        (
            cd /home/pi/opt
            curl http://openframeworks.cc/versions/v0.9.8/of_v0.9.8_linuxarmv6l_release.tar.gz | tar -xzf -
            mv of_v0.9.8_linuxarmv6l_release of
            #sudo of/scripts/linux/debian/install_dependencies.sh
        )
    fi
}

• run it $ sudo sh ./nsynth-setup.sh
• reboot $ sudo reboot

I try to generate the audio files but it fails at the command :

nsynth_generate --checkpoint_path=/home/user/magenta/magenta/models/nsynth/wavenet-ckpt/model.ckpt-200000 --source_path=/home/user/nsynth-pipeline/workdir/embeddings_batched/batch0 --save_path=/home/user/nsynth-pipeline/workdir/audio_output/batch0 --batch_size=512 --alsologtostderr --gpu_number=0

With the output log (it stops running after) :

WARNING:tensorflow:From /usr/local/lib/python2.7/dist-packages/magenta/models/nsynth/wavenet/masked.py:116: init (from tensorflow.python.ops.init_ops) is deprecated and will be removed in a future version.
Instructions for updating:
Use tf.initializers.variance_scaling instead with distribution=uniform to get equivalent behavior.

Can't understand why...

NOTE : I have all the files in embeddings_batched and embeddings_input folders, but just empty directories (batch0, ..., batch15) in audio_output

Since I can't find PEC11R-4315F-N0012-ND, I use PEC11R-4315F-S0012-ND as encoders(also works according to the instruction). However, the encoders have 5 pins with 3 on one side and 2 on the other, while there are only 3 cutouts on the board. I looked at the PCB image and found that the demo board has 3 pins as well. How do I deal with this? Can I just cut off the extra 2 pins of the encoders?

I turn on my nsynth; no midi plugged in; audio plugged in; and power;

oled screen turns on; works a little while; and turns off.

I tried debugging by running the app manually, and poking around, but could use some help as to where to poke next

I would like to send MIDI over USB from a Traktor Pro Remix Deck (in step sequence or one-shot mode) to be able to use the NSynth Super in a live environment while DJ'ing.

Is it possible to send MIDI over USB? Or do I have to use the 5-pin DIN port? If this is currently not possible, could it be added using an openFrameworks package?
I couldn't find anything about it in the docs, and would try myself, but I'm currently waiting for the PCB to arrive.

Resolved with:

$ sudo apt-get install libav-tools lame libsox-fmt-all

Docs need updating. See also #47.

I noticed in the hardware folder that display communicates via i2c, does that mean we could actually use a generic i2c OLED instead of the 20$ adafruit one? As long as we solder the 4 pins in the right place it should work, right?

More a question than an issue. I'm looking through the code trying to understand how the Multisampler works.

What does the parameter position (in some parts referred to as offset when multiplied with ncols) do? I can deduce it can be changed by the first potentiometer, but still can't wrap my head around how it changes the synth.

Is there anyone who is working on a plugin similar to the Ableton's one?

After building, installing, and successively testing the board with the Raspberry Pi 3 model B connected to a monitor, I shut down the system and tested the board in standalone mode. For some reason, the y-coordinate becomes “stuck” on the third row from the bottom, and by stuck, I mean that it’s quite difficult (but not impossible) to coax the grid indicator on the panel to move up or down to another y-coordinate by pressing on another row for a few seconds (the x-coordinate works fine).

On a whim, I plugged in the HDMI cable again, and the grid started working normally (and back to ‘sticky’ grid behavior when the cable is unplugged). I tried the 64GB image, but it had the same issue. It’s almost as if there’s an imaginary line on the 3rd-last row that make it difficult to cross when nothing is plugged into the HDMI port (unplugging or plugging in the mouse and keyboard had no effect).

Had anyone had the same odd grid behaviour? Could it have something to do with the model B and/or video driver? I double-checked the grid behaviour right to the IC pins with a scope and can’t see anything unusual between working and not working modes, which makes me think it’s software-related.

I've been going through the documentation to see how the Raspberry Pi needs to be mounted, but I couldn't find any pictures. Everything else, like the OLED, the enclosure and all the mechanical parts is very well documented (hats off!), but it seems to be missing for the R Pi.

Followed audio generation instructions and everything went fine until I got to executing nsynth_generate which, with or without arguments, fails with:

andrew@blast:/data/Liles/nsynth/open-nsynth-super/audio/workdir$ nsynth_generate
WARNING:tensorflow:From /usr/local/lib/python2.7/dist-packages/magenta/models/nsynth/wavenet/masked.py:116: init (from tensorflow.python.ops.init_ops) is deprecated and will be removed in a future version.
Instructions for updating:
Use tf.initializers.variance_scaling instead with distribution=uniform to get equivalent behavior.
Traceback (most recent call last):
File "/usr/local/bin/nsynth_generate", line 7, in
from magenta.models.nsynth.wavenet.nsynth_generate import console_entry_point
File "/usr/local/lib/python2.7/dist-packages/magenta/models/nsynth/wavenet/nsynth_generate.py", line 28, in
tf.app.flags.DEFINE_string("npy_only", False, "If True, use only .npy files.")
File "/usr/local/lib/python2.7/dist-packages/tensorflow/python/platform/flags.py", line 58, in wrapper
return original_function(*args, **kwargs)
File "/usr/local/lib/python2.7/dist-packages/absl/flags/_defines.py", line 241, in DEFINE_string
DEFINE(parser, name, default, help, flag_values, serializer, **args)
File "/usr/local/lib/python2.7/dist-packages/absl/flags/_defines.py", line 81, in DEFINE
DEFINE_flag(_flag.Flag(parser, serializer, name, default, help, **args),
File "/usr/local/lib/python2.7/dist-packages/absl/flags/_flag.py", line 107, in init
self._set_default(default)
File "/usr/local/lib/python2.7/dist-packages/absl/flags/_flag.py", line 196, in _set_default
self.default = self._parse(value)
File "/usr/local/lib/python2.7/dist-packages/absl/flags/_flag.py", line 169, in _parse
'flag --%s=%s: %s' % (self.name, argument, e))
absl.flags._exceptions.IllegalFlagValueError: flag --npy_only=False: flag value must be a string, found "<type 'bool'>"

Any suggestions as to a remedy would be much appreciated!

Hi experts,
OK built my unit,,,,

I have loaded the SD card with the image , booted and get presented with this image of lots of colours but it goes no further , any advice please

I'm on a waitlist to order a PCB to build an NSynth Super at the moment so it may be a bit before I'm able to test this, but I'm curious whether MPE is supported by this?

If not familiar, this is a new specification recently adopted by MIDI and allows multidimensional controllers such as the LinnStrument and Seaboard to add pitch bends, vibrato, etc to individual notes rather than the all of the notes played simultaneously.

If not currently an option, I'd love to see it in a future addition to the hardware/software 👍

Here is a link to MIDI's release on the standard for more info:
https://www.midi.org/articles/midi-polyphonic-expression-mpe

Is NSynth Super polyphonic synth?

Can I run sound creating pipeline on a server without a GPU? I understand that the process will be long, but I do not have access to a server with a GPU.

Thanks for making this and open sourcing it!

My Open NSynth Super plays only a fraction of the notes that I play on my MIDI keyboard. Most of them are missed and do not result in any sound. Does anybody else have the same issue? Any ideas what the problem could be?
Also, selecting some of the sounds like, for example, "El. Tom", "Snare", or "Vox. Hit" does not result in any audible sound whatsoever. Any clues?

Hi !

I'm wondering if it could be possible to put some potentiometers instead of the 4 rotary encoders in the corners (the good reference is impossible to find on the internet and I had the one with 2 more arms -or could I cut them ?- )

Thanks !

Hi! Great project! The build was really fun, hope I didn't mess it up. 😜So I wrote 'onss_1.1_full.img' to a 64GB card and tried to follow the instructions in the firmware installation section of the make guide. It tells me:

Next, navigate to the firmware directory and run the install command:

$ cd /home/pi/open-nsynth-super-master/firmware/src
$ make install

Problem is I can't find the directory 'open-nsynth-super-master' anywhere on the card. I found 'open-nsynth-super' -- which leads down the same hierarchy of directories -- but when I 'make install' from there, I get an error: "could not intialize the debug port."

Any thoughts? Thanks in advance!!

The Midi thread currently misinterprets "system real-time" messages (which may be inserted into the midi byte stream at any point) as malformed data and exits. This makes nsynth unusable in a production environment where midi clock messages are used to keep devices in sync.

ref: http://www.lim.di.unimi.it/IEEE/MIDI/SOT3.HTM "Real Time messages can be sent at any time and may be inserted anywhere in a MIDI data stream, including between Status and Data bytes of any other MIDI messages."

I am trying to build the OS in a 64GB SD card under Ubuntu 17.10 but get errors when I try to access the card:

fdisk: cannot open /dev/nvme0n1: Permission denied
fdisk: cannot open /dev/sda: Permission denied

I looked for solutions but doesn't get much help. One solution suggests I add myself into the group that denies my permission, but the group "nvme0n1" doesn't even exist. Does anyone know how to solve the issue? Moreover, is there an easy way to build the OS in the SD card for Windows user? I have a Surface Book 2 and the Ubuntu installed in my laptop is very unstable.

I am so close to getting this working. Even though my display wont power on, I can use the HDMI port on the pi to check everything out. Nsynth is running, and all the knobs + touchpad work. When I run the speaker-test, however, I get an error:

Playback open error: -16, Device or resource busy

Any ideas on what could be causing this?

Thanks

I'm desperate for some help. I completed build a few days ago and have been unable to resolve the issue on my own. I cannot get any sounds to be made with my synth. I ran a sound test through a command and was able to hear noise out of the headphone jack. That leads me to believe that no Midi signal is reaching the device. I tried using the 64 and 16 GB preconfigured images. With both, everything else appears to function just by plugging the imaged card into my system. The screen functions and shows an expected response when each dial is turned. On the Midi connection, I have tried plugging in a keyboard with 5 pin end on each side (no USB involved) and also tried playing some notes out of a couple of DAWs installed on my PC and connecting with a USB to Midi cable. Are there any diagnostics that I can run to see if a signal is even being received? Is there something additional I needed to do in first setup of the software? Perhaps something else basic that I missed? I don't have much knowledge in this area, so I won't be offended by any suggestions. Thanks in advance!

How can I close the GUI OLED elements to use the CLI on boot? ps, and top don't allow me to see all processes as the GUI is overlayed on top of the CLI with a large surrounding black box.

After having played the Open NSynth Super for minutes, the volume starts fading out gradually (over perhaps 30 seconds or so) until I don't hear anything anymore. Adjusting the volume pot has no effect as does adjusting the volume in the ALSA Mixer. When terminating the Open NSynth Super process and playing some audio file via the aplay command, it plays perfectly fine. When I restart Open NSynth Super afterwards it works fine again (for another couple of minutes). Does anybody have any idea what might be going on?

Hi !

I'm trying to install the firmware on the raspberry.
Nothing happens when I run "sudo mount -o remount,rw /"
And if I try to run "cd home/pi/open-nsynth-super-master/firmware/src" then it says "No such file or directory".

I tried already to set the image on the micro SD again, but nothing changed.

Does anyone know ?

Run on a 2x GPU system with 4x instruments per corner. Completes fine with one batch and not the other. Guess the error is misleading...

@jesseengel, wondering if you've seen this? (and doesn't seem like anyone else is watching this repo!)

//

(magenta) andrew@blast:/data/Liles/nsynth/open-nsynth-super/audio/workdir$
(magenta) andrew@blast:/data/Liles/nsynth/open-nsynth-super/audio/workdir$ python 05_clean_files.py 1
Normalising input WAVs
2%|█▊ | 992/57784 [03:26<3:16:57, 4.81it/s]100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████| 57784/57784 [3:24:43<00:00, 4.70it/s]
Converting to WAV
100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████| 104431/104431 [41:26<00:00, 42.00it/s]
(magenta) andrew@blast:/data/Liles/nsynth/open-nsynth-super/audio/workdir$
(magenta) andrew@blast:/data/Liles/nsynth/open-nsynth-super/audio/workdir$
(magenta) andrew@blast:/data/Liles/nsynth/open-nsynth-super/audio/workdir$
(magenta) andrew@blast:/data/Liles/nsynth/open-nsynth-super/audio/workdir$
(magenta) andrew@blast:/data/Liles/nsynth/open-nsynth-super/audio/workdir$
(magenta) andrew@blast:/data/Liles/nsynth/open-nsynth-super/audio/workdir$ python 05_clean_files.py 0
Normalising input WAVs
27%|█████████████████████████████▊ | 15537/57784 [58:29<2:39:03, 4.43it/s]sox WARN gain: gain clipped 63999 samples; decrease volume?
sox WARN dither: dither clipped 56045 samples; decrease volume?
28%|███████████████████████████████ | 16454/57784 [1:01:56<2:35:35, 4.43it/s]sox WARN gain: gain clipped 63902 samples; decrease volume?
sox WARN dither: dither clipped 56048 samples; decrease volume?
70%|███████████████████████████████████████████████████████████████████████████▉ | 40228/57784 [2:31:36<1:06:09, 4.42it/s]sox WARN gain: gain clipped 64000 samples; decrease volume?
sox WARN dither: dither clipped 55997 samples; decrease volume?
81%|█████████████████████████████████████████████████████████████████████████████████████████▌ | 46647/57784 [2:55:49<41:58, 4.42it/s]Traceback (most recent call last):
File "05_clean_files.py", line 46, in
audio, sr = librosa.core.load(os.path.join(original_path, fpath), sr=16000)
File "/data/Liles/nsynth/miniconda2/envs/magenta/lib/python2.7/site-packages/librosa/core/audio.py", line 112, in load
with audioread.audio_open(os.path.realpath(path)) as input_file:
File "/data/Liles/nsynth/miniconda2/envs/magenta/lib/python2.7/site-packages/audioread/init.py", line 116, in audio_open
raise NoBackendError()
audioread.NoBackendError

I cloned the repo and opened the dxf file in DWG TrueView.

I get the following error when doing so:

The Midi thread currently misinterprets "system real-time" messages (which may be inserted into the midi byte stream at any point) as malformed data and exits. This makes nsynth unusable in a production environment where midi clock messages are used to keep devices in sync.

ref: http://www.lim.di.unimi.it/IEEE/MIDI/SOT3.HTM "Real Time messages can be sent at any time and may be inserted anywhere in a MIDI data stream, including between Status and Data bytes of any other MIDI messages."

Hi there, I can see a massive difference between this model and the one posted by Google which is the touchscreen is missing from this design. Is there a way to include the touchscreen from adafruit instead of the capacitative pad?

The git tree on sf.net for openocd is down, reference here. The Nsynth bash file would hang on the "Cloning into 'openocd' step. I had to modify nsynth-setup.sh to pull from the secondary (official) git repo for openOCD:

git://repo.or.cz/openocd.git

(This is for future info... not really a bug, but in case someone else runs into the same issue in the meantime)

Looking to port some of these ideas to other platforms (i.e. iOS). Downloaded the "lite" audio binary files and the respective settings.json.

Looked through some of the source code, but got a little confused in the process. Can someone walk me through the process of using the binary audio files to generate a sound wave given an X, Y pair on the touchscreen plane?

Thanks.

Currently the capacitive touch grid is mapped to 0-10 or a point per touch IC sensor on each axis.

Is it possible to map the cap touch grid to a higher resolution?
Looking at the datasheet for the ICs suggests it is possible for much higher resolution with triangular trace designs.

I have tried resoldering the jumpers and the pins but nothing is powering it. Nsynth makes noise when plugged into my MPC. So I know that at least works. Any ideas? Is there a firmware for the OLED that I am maybe missing?

My PCB (ordered on Tindle) doesn't have the right holes (not like on the picture) to put the 3,5mm audio jack !

Does anyone had the same problem ?

Thanks !

running $ python 01_compute_input_embeddings.py from ~/open-nsynth-super/audio/workdir to test the pipeline gives me an OSError: [Errno 2] No such file or directory

I split up the subprocess.call to get a more specific line, though I don't think that worked (it breaks on the last line regardless).

Traceback:

Traceback (most recent call last):
  File "01_compute_input_embeddings.py", line 31, in <module>
    subprocess.call(["nsynth_save_embeddings", "--checkpoint_path=%s/magenta/models/nsynth/wavenet-ckpt/model.ckpt-200000" % settings['magenta_dir'], "--source_path=%s/audio_input" % source_dir, "--batch_size=64", "--save_path=%s/embeddings_input" % source_dir])
  File "/usr/lib/python2.7/subprocess.py", line 523, in call
    return Popen(*popenargs, **kwargs).wait()
  File "/usr/lib/python2.7/subprocess.py", line 711, in __init__
    errread, errwrite)
  File "/usr/lib/python2.7/subprocess.py", line 1343, in _execute_child
    raise child_exception
OSError: [Errno 2] No such file or directory

all folders exist and are readable, writeable and executable by current user

I think it might have to do something with the wavenet-ckpt folder which contains checkpoint, model.ckpt-200000.data-00000-of-00001, model.ckpt-200000.index and model.ckpt-200000.meta but no model.ckpt-200000

Hello :)
What exactly do you mean by "obtain"?

Hi there,
I'm building one of these and following along with the readme: https://github.com/googlecreativelab/open-nsynth-super/blob/master/linux/readme.md

I'm having an issue on Step 6. All other steps were completed without error. However, on the Raspberry Pi when executing
$ bash nsynth-setup.sh

I receive four errors:
line 11: $'\r': command not found
line 12: $'\r: command not found
line 13: syntax error near unexpected token '$' {\r''
line 13: 'setup_audio() {

and then it returns...

One thing that it different from the instructions is that i've copied over the directory structure open-nsynth-super/ and not open-nsynth-super-master/ as I've been unable to locate that...?

Any help would be most appreciated.

I am trying to run the pipeline on a OSX system and running into this error on the 4th Generate audio
step:

File "/anaconda3/lib/python3.6/site-packages/tensorflow/python/client/session.py", line 1111, in _run
str(subfeed_t.get_shape())))
ValueError: Cannot feed value of shape (36, 125, 16) for Tensor 'encoding:0', which has shape '(36, 16)'

I've also tried this on a windows 10 system and running into the same error.

Any ideas?

Cheers!

The table of reference codes for the surface mount components printed on the back of the pcb contains every component except the diode, which is listed as D1. There's an unlisted connection labeled D1 to the right of the display. The below image shows the D1 connection at the top right and the list on the far right.

it seems that the bottom and top paste files arent included in the renders, is there a reason why or is it a mistake?

Once the device is up&running as a standalone synth, is there any way or guideline to gracefully invoke the raspberry shutdown from the hardware user interface?

When trying to use sudo wpa_cli it hangs 'Could not connect to wpa_suplicant:wlan0 - re-trying`. When I trace the wpa_cli I see it repeating that it's a read-only file system.

Do we need to set up an new wpa_supplicant.conf someplace outside the locked image?

Was able to connect by remounting and editing the wpa config. Maybe instruction change?

Are you likely to publish schematics for the model in the demo video with the colour touch LCD display? Although I appreciate the model here is designed to be simpler to build, having such a large display makes the thing look even more awesome, so if it weren't too expensive to do I'd love to give that model a try.

Cheers,
Rob