Skip to main content

ACES metric for evaluating automated audio captioning models based on the semantics of sounds

Project description

ACES

This is the repository of Audio Captioning Evaluation on Semantics of Sound (ACES).

In here you will find the instructions how to train an ACES model and calculate statistics.

Installation

pip install aces-metric

Usage

The candidates can be a list, the references can be a list or a list of lists.

from aces import get_aces_score
candidates = ["a bunch of birds are singing"]
references = ["birds are chirping and singing loudly in the forest"]
score = get_aces_score(candidates, references, average=True)

Semantics of sounds

To get an output of classes of semantic groups from a caption:

from transformers import pipeline
pipe = pipeline("token-classification", "gijs/aces-roberta-13", aggregation_strategy="simple")
pipe("Bird chirps in the tree while a car hums")

Evaluation

All the code that is used to evaluate different models for the research paper can be found in the evaluation folder on the github. Particularly, the model evaluation can be found in evaluation/eval.py, and information about the FENSE experiment can be found in evaluation/fense_experiment/main.py.

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

aces-metric-0.0.4.tar.gz (12.5 kB view details)

Uploaded Source

Built Distribution

aces_metric-0.0.4-py3-none-any.whl (13.2 kB view details)

Uploaded Python 3

File details

Details for the file aces-metric-0.0.4.tar.gz.

File metadata

  • Download URL: aces-metric-0.0.4.tar.gz
  • Upload date:
  • Size: 12.5 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/4.0.2 CPython/3.9.18

File hashes

Hashes for aces-metric-0.0.4.tar.gz
Algorithm Hash digest
SHA256 9b8beea5760b0f5e509c9d1aeb246b4aecdf6799e7c3218a09cb186578c1c7c8
MD5 83985e4a3af9ffd995473ecf3a492dbf
BLAKE2b-256 a35ea5178190b36aa196bbd983450495457a8a96a203e479d6443d9b952e192b

See more details on using hashes here.

File details

Details for the file aces_metric-0.0.4-py3-none-any.whl.

File metadata

  • Download URL: aces_metric-0.0.4-py3-none-any.whl
  • Upload date:
  • Size: 13.2 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/4.0.2 CPython/3.9.18

File hashes

Hashes for aces_metric-0.0.4-py3-none-any.whl
Algorithm Hash digest
SHA256 8b1526374996d4d72d08853c6b177094619b9e1f641daafe311ae9134c86e423
MD5 1b1a9fabc3bd1632019d830bd6b99340
BLAKE2b-256 6aa20e4bb6672b3386b88b397785e9b507367ce930b05f30b0c77e3080cf289e

See more details on using hashes here.

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page