tensorflow / io

Dataset, streaming, and file system extensions maintained by TensorFlow SIG-IO

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

tfio M1 0.34.0 arm64: AudioIOTensor has no attribute 'io_audio_readable_init'

yingding opened this issue · comments

I got an error no attribute io_audio_readable_init on M1 Max MBP with Python 3.10.

pip install tensorflow-io==0.34.0 tensorflow==2.13.0 tensorflow-metal==1.0.1

Run the tfio example code:

import tensorflow_io as tfio
audio = tfio.audio.AudioIOTensor('gs://cloud-samples-tests/speech/brooklyn.flac')

Then got the error no attribute 'io_audio_readable_init' for AudioIOTensor:

File [~/VENV/tfio3.10/lib/python3.10/site-packages/tensorflow_io/python/ops/audio_ops.py:671](https://file+.vscode-resource.vscode-cdn.net/Users/yingding/VCS/github/ml/tf-every-day/08Audio/~/VENV/tfio3.10/lib/python3.10/site-packages/tensorflow_io/python/ops/audio_ops.py:671), in AudioIOTensor.__init__(self, filename, dtype)
    669 if not tf.executing_eagerly():
    670     assert dtype is not None, "dtype must be provided in graph mode"
--> 671 resource = core_ops.io_audio_readable_init(filename)
    672 if tf.executing_eagerly():
    673     shape, dtype, rate = core_ops.io_audio_readable_spec(resource)

File [~/VENV/tfio3.10/lib/python3.10/site-packages/tensorflow_io/python/ops/__init__.py:88](https://file+.vscode-resource.vscode-cdn.net/Users/yingding/VCS/github/ml/tf-every-day/08Audio/~/VENV/tfio3.10/lib/python3.10/site-packages/tensorflow_io/python/ops/__init__.py:88), in LazyLoader.__getattr__(self, attrb)
     87 def __getattr__(self, attrb):
---> 88     return getattr(self._load(), attrb)

AttributeError: module '56c055b4afaba95af0256db7c72edd0d967e1981' has no attribute 'io_audio_readable_init'

The root cause might be the same, which also is causing #1855

I have the same problem using same tf, tf-io and tf-metal versions as @yingding
MacOS: 13.4
Proc: M2 Pro

same issue. can not use Tensorflow.io on Mac M1

I have the same issue. Same tf and tfio versions and M1 system.

same issue on mbp with m1 pro using python 3.9, tfio 0.34.0, and tf 2.13.0

Same issue for me.

Any updates on this?, have tried installing tensorflow io from pypi and downloading from source following this guide,
i get same issue, look at source code and i can see the audio ops there, but not working.

recent setup attempt was:
python 3.11
tf: 2.15
tfio:0.36

as suggested here

same issue on m3 with tf 2.15.0 and tfio 0.36.0
any workaround ?

same issue on m3 with tf 2.15.0 and tfio 0.36.0 any workaround ?

I have similar issues with the same versions and also with tfio 0.34.0. I use Mac M1. I would be happy about a quick workaround.

Error AttributeError: module '77ab628d7ad4acaa62f6bde524b9d631895821c9' has no attribute 'io_audio_resample'

Same here with tfio 0.34.0, Tensorflow 2.13 and tensorflow-metal 1.0.0

Folks at Google also use MacBooks for dev... I'm wondering how come they don't have this problem?

i have this issue too usign tfio.audio.resample with m1Max
as im reading this i got that this issue just happend on apple silicon
that means this library works on regular windows systems??

I got the same error on the Apple M3 chip when using the audio module. It looks like the culprit is libtensorflow_io.so that is missing some functions. My environment:
python 3.10.14
tensorflow 2.15.0
tensorflow-io 0.36.0

The core_ops tries to access the io_audio_resample function, which is backed by the libtensorflow_io.so. Here is the initialization code in tensorflow io package: core_ops = LazyLoader("core_ops", "libtensorflow_io.so"). The core_ops did not load as many functions as the one from the Apple Intel chip. I compared the two runtimes side by side.

File "[.../venv/lib/python3.10/site-packages/tensorflow_io/python/ops/audio_ops.py", line 465] in f
        return core_ops.io_audio_resample(
File "[.../venv/lib/python3.10/site-packages/tensorflow_io/python/ops/__init__.py", line 88] in __getattr__
        return getattr(self._load(), attrb)

AttributeError: module '77ab628d7ad4acaa62f6bde524b9d631895821c9' has no attribute 'io_audio_resample'

same here with Tensorflow 2.16.1, Tensorflow-io 0.37.0 on Apple M1