Problem with deepspeech training (latest version), not recognised as a module when making csv/wav files with bin/import

DeepSpeech
, ,
#1

Hey I’m using google collab for training the lastest version of Deepspeech on a Spanish training set and I have some problem at the data preprocessing part here is the GitHub gist of the full code I have done . https://gist.github.com/aszi09/96bad06114310f2cf5b517df37352e16

It’s basically an error on the last part saying that there is no deepspeech_training module. (included in the gist)

However there are other errors I have ran into which I have absolutely no idea if are important or mean anything in terms of training and future inference. I can provide other info as well , it’s just that I’m new to this.

Also, I have another training going on another dataset with an earlier version which works (0.7.4) after 50 minutes it still hasn’t completed an epoch, I didn’t change any parameters when running so I’m unsure as to what the issue may be. (unsupported CUDA version may be the problem as Google collab uses 10.1)

Thanks in advance

#2

This is not clear. Depends on when, how you perform things.

So you are using master, please dont if you are not ready to experience some failures.

Read the docs: 0.9.0a7 was published under mvs_ctcdecoder package, we reverted that but have not yet published a new package under the new old name. Please either stick to v0.8.2 tag for a stable release or v0.9.0-alpha.7 for master-that-works.

Without more information, we can’t help.

#3

i will try with the v0.8.2 release thanks, for the other problem I will write back if I see any error I have with v0.8.2 release . Thanks

#4

FTR other community have had success using https://github.com/Common-Voice/commonvoice-fr/blob/master/DeepSpeech/Dockerfile.train but I can’t tell if it works on Google Collab.

#5

What is supposed to be the output after the “make Dockerfile.train” command? In the lastest version I got “No rules … stopping” or along the lines of that but it’s inside the gist somewhere. However, with the 0.7.4 version of deep speech I did manage to start training and had a different output for “make Dockerfile.train”. I will try with the version you suggested and will get back to you if something is misbehaving . Thanks in advance

#6

Have you read and understood CONTRIBUTING.md ? There’s no such requirement.

Please stick to our documantion and not some “gist somewhere”. We can’t provide support on that.

Again, you are mixing things.

#7

Hey i tried setting up training with the version u suggested v0.8.2. First problem I had , which Im guessing is related to pip is this error message after !pip3 install --upgrade -e.

Obtaining file:///content/DeepSpeech
Requirement already satisfied, skipping upgrade: numpy in /usr/local/lib/python3.6/dist-packages (from deepspeech-training==0.8.2) (1.18.5)
Requirement already satisfied, skipping upgrade: progressbar2 in /usr/local/lib/python3.6/dist-packages (from deepspeech-training==0.8.2) (3.38.0)
Requirement already satisfied, skipping upgrade: six in /usr/local/lib/python3.6/dist-packages (from deepspeech-training==0.8.2) (1.15.0)
Collecting pyxdg
  Using cached pyxdg-0.26-py2.py3-none-any.whl (40 kB)
Collecting attrdict
  Using cached attrdict-2.0.1-py2.py3-none-any.whl (9.9 kB)
Requirement already satisfied, skipping upgrade: absl-py in /usr/local/lib/python3.6/dist-packages (from deepspeech-training==0.8.2) (0.8.1)
Collecting semver
  Using cached semver-2.10.2-py2.py3-none-any.whl (12 kB)
Collecting opuslib==2.0.0
  Using cached opuslib-2.0.0.tar.gz (7.3 kB)
Collecting optuna
  Using cached optuna-2.0.0.tar.gz (226 kB)
Collecting sox
  Using cached sox-1.4.0-py2.py3-none-any.whl (39 kB)
Requirement already satisfied, skipping upgrade: bs4 in /usr/local/lib/python3.6/dist-packages (from deepspeech-training==0.8.2) (0.0.1)
Requirement already satisfied, skipping upgrade: pandas in /usr/local/lib/python3.6/dist-packages (from deepspeech-training==0.8.2) (1.0.5)
Requirement already satisfied, skipping upgrade: requests in /usr/local/lib/python3.6/dist-packages (from deepspeech-training==0.8.2) (2.23.0)
Collecting numba==0.47.0
  Using cached numba-0.47.0-cp36-cp36m-manylinux1_x86_64.whl (3.7 MB)
Requirement already satisfied, skipping upgrade: llvmlite==0.31.0 in /usr/local/lib/python3.6/dist-packages (from deepspeech-training==0.8.2) (0.31.0)
Requirement already satisfied, skipping upgrade: librosa in /usr/local/lib/python3.6/dist-packages (from deepspeech-training==0.8.2) (0.6.3)
Collecting soundfile
  Using cached SoundFile-0.10.3.post1-py2.py3-none-any.whl (21 kB)
Collecting ds_ctcdecoder==0.8.2
  Using cached ds_ctcdecoder-0.8.2-cp36-cp36m-manylinux1_x86_64.whl (2.0 MB)
Collecting tensorflow==1.15.2
  Using cached tensorflow-1.15.2-cp36-cp36m-manylinux2010_x86_64.whl (110.5 MB)
Requirement already satisfied, skipping upgrade: python-utils>=2.3.0 in /usr/local/lib/python3.6/dist-packages (from progressbar2->deepspeech-training==0.8.2) (2.4.0)
Collecting alembic
  Using cached alembic-1.4.2.tar.gz (1.1 MB)
  Installing build dependencies ... done
  Getting requirements to build wheel ... done
ERROR: Exception:
Traceback (most recent call last):
  File "/usr/local/lib/python3.6/dist-packages/pip/_internal/cli/base_command.py", line 186, in _main
    status = self.run(options, args)
  File "/usr/local/lib/python3.6/dist-packages/pip/_internal/commands/install.py", line 331, in run
    resolver.resolve(requirement_set)
  File "/usr/local/lib/python3.6/dist-packages/pip/_internal/legacy_resolve.py", line 177, in resolve
    discovered_reqs.extend(self._resolve_one(requirement_set, req))
  File "/usr/local/lib/python3.6/dist-packages/pip/_internal/legacy_resolve.py", line 333, in _resolve_one
    abstract_dist = self._get_abstract_dist_for(req_to_install)
  File "/usr/local/lib/python3.6/dist-packages/pip/_internal/legacy_resolve.py", line 282, in _get_abstract_dist_for
    abstract_dist = self.preparer.prepare_linked_requirement(req)
  File "/usr/local/lib/python3.6/dist-packages/pip/_internal/operations/prepare.py", line 516, in prepare_linked_requirement
    req, self.req_tracker, self.finder, self.build_isolation,
  File "/usr/local/lib/python3.6/dist-packages/pip/_internal/operations/prepare.py", line 95, in _get_prepared_distribution
    abstract_dist.prepare_distribution_metadata(finder, build_isolation)
  File "/usr/local/lib/python3.6/dist-packages/pip/_internal/distributions/sdist.py", line 38, in prepare_distribution_metadata
    self._setup_isolation(finder)
  File "/usr/local/lib/python3.6/dist-packages/pip/_internal/distributions/sdist.py", line 96, in _setup_isolation
    reqs = backend.get_requires_for_build_wheel()
  File "/usr/local/lib/python3.6/dist-packages/pip/_vendor/pep517/wrappers.py", line 152, in get_requires_for_build_wheel
    'config_settings': config_settings
  File "/usr/local/lib/python3.6/dist-packages/pip/_vendor/pep517/wrappers.py", line 255, in _call_hook
    raise BackendUnavailable(data.get('traceback', ''))
pip._vendor.pep517.wrappers.BackendUnavailable: Traceback (most recent call last):
  File "/usr/local/lib/python3.6/dist-packages/pip/_vendor/pep517/_in_process.py", line 63, in _build_backend
    obj = import_module(mod_path)
  File "/usr/lib/python3.6/importlib/__init__.py", line 126, in import_module
    return _bootstrap._gcd_import(name[level:], package, level)
  File "<frozen importlib._bootstrap>", line 994, in _gcd_import
  File "<frozen importlib._bootstrap>", line 971, in _find_and_load
  File "<frozen importlib._bootstrap>", line 941, in _find_and_load_unlocked
  File "<frozen importlib._bootstrap>", line 219, in _call_with_frames_removed
  File "<frozen importlib._bootstrap>", line 994, in _gcd_import
  File "<frozen importlib._bootstrap>", line 971, in _find_and_load
  File "<frozen importlib._bootstrap>", line 955, in _find_and_load_unlocked
  File "<frozen importlib._bootstrap>", line 665, in _load_unlocked
  File "<frozen importlib._bootstrap_external>", line 678, in exec_module
  File "<frozen importlib._bootstrap>", line 219, in _call_with_frames_removed
  File "/usr/local/lib/python3.6/dist-packages/setuptools/__init__.py", line 5, in <module>
    import distutils.core
  File "/tmp/pip-build-env-mmrfxsqo/overlay/lib/python3.6/site-packages/_distutils_hack/__init__.py", line 82, in create_module
    return importlib.import_module('._distutils', 'setuptools')
  File "/usr/lib/python3.6/importlib/__init__.py", line 126, in import_module
    return _bootstrap._gcd_import(name[level:], package, level)
ModuleNotFoundError: No module named 'setuptools._distutils'

After this, I tried continuing the process described in the documentation of v0.8.2 release but when I ran !bin/import_cv2.py I got this

/bin/sh: 1: sox: not found
SoX could not be found!

    If you do not have SoX, proceed here:
     - - - http://sox.sourceforge.net/ - - -

    If you do (or think that you should) have SoX, double-check your
    path variables.
    
Traceback (most recent call last):
  File "bin/import_cv2.py", line 18, in <module>
    from deepspeech_training.util.downloader import SIMPLE_BAR
ModuleNotFoundError: No module named 'deepspeech_training'

The whole process I did is inside this gist.

Thanks for your response in advance .

#8

no idea why, but this is obviously not a bug on our side

install the required deps?

#9

This is weird, it also breaks on Docker train CI: https://github.com/mozilla/DeepSpeech/issues/3295 but not on other places. I have no idea what is going on …

#10

Im quite new so I don’t exactly know what that means, but if you need me to put in any lines that will help with resolving the issue tell me.

#11

As I said, it looks like it also happens on our CI, but I don’t have time to look at that right now …

#12

@arpi.aszalos Looks like it’s a Python level issue: https://github.com/pypa/setuptools/issues/2353

#13

This error also seems to be a side-effect of the linked Pypa/Setuptools error, and if you pip install --upgrade . instead of pip install --upgrade -e . it should work. You might need also setuptools==50.0.0 as well?

#14

okay i will try with the deepspeech v0.8.2 version. Thank you very much for helping

#15

I tried with the v0.8.2 version of deepspeech and used !pip3 install --upgrade .
The error doesn’t occur anymore however this message was shown :

Processing /content/DeepSpeech
Requirement already satisfied, skipping upgrade: numpy in /usr/local/lib/python3.6/dist-packages (from deepspeech-training==0.9.0a7) (1.18.5)
Requirement already satisfied, skipping upgrade: progressbar2 in /usr/local/lib/python3.6/dist-packages (from deepspeech-training==0.9.0a7) (3.38.0)
Requirement already satisfied, skipping upgrade: six in /usr/local/lib/python3.6/dist-packages (from deepspeech-training==0.9.0a7) (1.15.0)
Collecting pyxdg
  Downloading pyxdg-0.26-py2.py3-none-any.whl (40 kB)
     |████████████████████████████████| 40 kB 3.4 MB/s 
Collecting attrdict
  Downloading attrdict-2.0.1-py2.py3-none-any.whl (9.9 kB)
Requirement already satisfied, skipping upgrade: absl-py in /usr/local/lib/python3.6/dist-packages (from deepspeech-training==0.9.0a7) (0.8.1)
Collecting semver
  Downloading semver-2.10.2-py2.py3-none-any.whl (12 kB)
Collecting opuslib==2.0.0
  Downloading opuslib-2.0.0.tar.gz (7.3 kB)
Collecting optuna
  Downloading optuna-2.0.0.tar.gz (226 kB)
     |████████████████████████████████| 226 kB 12.9 MB/s 
Collecting sox
  Downloading sox-1.4.0-py2.py3-none-any.whl (39 kB)
Requirement already satisfied, skipping upgrade: bs4 in /usr/local/lib/python3.6/dist-packages (from deepspeech-training==0.9.0a7) (0.0.1)
Requirement already satisfied, skipping upgrade: pandas in /usr/local/lib/python3.6/dist-packages (from deepspeech-training==0.9.0a7) (1.0.5)
Requirement already satisfied, skipping upgrade: requests in /usr/local/lib/python3.6/dist-packages (from deepspeech-training==0.9.0a7) (2.23.0)
Collecting numba==0.47.0
  Downloading numba-0.47.0-cp36-cp36m-manylinux1_x86_64.whl (3.7 MB)
     |████████████████████████████████| 3.7 MB 12.0 MB/s 
Requirement already satisfied, skipping upgrade: llvmlite==0.31.0 in /usr/local/lib/python3.6/dist-packages (from deepspeech-training==0.9.0a7) (0.31.0)
Requirement already satisfied, skipping upgrade: librosa in /usr/local/lib/python3.6/dist-packages (from deepspeech-training==0.9.0a7) (0.6.3)
Collecting soundfile
  Downloading SoundFile-0.10.3.post1-py2.py3-none-any.whl (21 kB)
ERROR: Could not find a version that satisfies the requirement ds_ctcdecoder==0.9.0-alpha.7 (from deepspeech-training==0.9.0a7) (from versions: 0.6.1, 0.7.0, 0.7.1, 0.7.3, 0.7.4, 0.8.0a3, 0.8.0a4, 0.8.0a5, 0.8.0a6, 0.8.0a7, 0.8.0a8, 0.8.0, 0.8.1, 0.8.2, 0.9.0a0, 0.9.0a1, 0.9.0a2, 0.9.0a3, 0.9.0a4, 0.9.0a5)
ERROR: No matching distribution found for ds_ctcdecoder==0.9.0-alpha.7 (from deepspeech-training==0.9.0a7)

Furthermore, when I try to **!bin/import_cv2.py ** I get this still

Traceback (most recent call last):
File “bin/import_cv2.py”, line 18, in
from deepspeech_training.util.downloader import SIMPLE_BAR
ModuleNotFoundError: No module named ‘deepspeech_training’

Here is the gist of the process I did

Thanks in advance!

#16

You can’t have that error on v0.8.2 …

Please understand it’s upstream tooling that is broken, I can’t help you.

#17

Which is meaningless since the previous install failed so deepspeech_training is not installed.

#18

yes sorry didnt include --branch v0.8.2 in cloning . thanks for helping. I will see if it works still

#19

I got it to import thank you so much.
I did have to change setuptools==50.0.0 and use !pip3 install --upgrade .

#20

Thanks, so it confirms there’s something bad with this release. Given the amount of people reporting issues against it in the past hours :confused: