Detect Problems in Audio Classification Datasets

Use the sliceguard library and Spotlight to quickly detect problem clusters that can cause issues when training your audio classification model. Shows problems such as:

1. Label Inconsistencies
2. Outliers and Errors
3. Audio-specific Issues (e.g. clipping)

Generally it will show you everything that is hard to learn for a classification model. All you need is a dataframe containing paths to audio files and labels.

First install the dependencies:

pip install renumics-spotlight sliceguard[all] scikit-learn

Then run the following code to detect problematic clusters:

# The Imports
from renumics import spotlight
from sliceguard import SliceGuard
from sliceguard.data import from_huggingface
from sklearn.metrics import accuracy_score

# Load an Example Dataset as DataFrame
df = from_huggingface("renumics/emodb")

# DataFrame Format:
# +---------------------+---------+
# |        audio        | emotion |
# +---------------------+---------+
# | /path/to/audio1.wav | joy     |
# | /path/to/audio2.wav | anger   |
# | /path/to/audio3.wav | joy     |
# | ...                 |         |
# +---------------------+---------+

# Detect Issues Using sliceguard
sg = SliceGuard()
issues = sg.find_issues(df, features=["audio"], y="emotion", metric=accuracy_score)
report_df, spotlight_data_issues, spotlight_dtypes, spotlight_layout = sg.report(
    no_browser=True
)

# Visualize Detected Issues in Spotlight:
spotlight.show(
    report_df,
    dtype=spotlight_dtypes,
    issues=spotlight_data_issues,
    layout=spotlight_layout,
)