# marl/crepe

CREPE: A Convolutional REpresentation for Pitch Estimation -- pre-trained model (ICASSP 2018)

Repository: https://github.com/marl/crepe
Canonical: https://ross.abutalabs.com/products/crepe
Homepage: https://marl.github.io/crepe/
Language: Python
License: MIT
License Family: permissive
Topics: pitch-estimation, tensorflow, keras, music-information-retrieval
Last push: 2024-08-19T18:12:09+00:00

## Health v2 (maintenance only)
Score: 23/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 0, release rhythm 8, longevity 100
- inputs: {"age_days": 3181, "days_push": 744, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1411, forks 186 (observed 2026-08-28T04:04:38.943085+00:00)

## What it is
CREPE is a monophonic pitch tracker based on a deep convolutional neural network that operates directly on time-domain audio waveforms, distributed as a Python package with a pre-trained TensorFlow/Keras model. It provides both a command-line utility and a Python API that output fundamental frequency (f0) and voicing confidence estimates over time.

## Use cases
- estimate the pitch of a wav file
- extract fundamental frequency contours from monophonic audio
- track pitch of singing voice recordings
- get voicing confidence scores for audio frames
- compare pitch trackers like pYIN and SWIPE
- batch process audio files for f0 analysis from the command line

## When to choose
- you need accurate monophonic pitch estimation on 16 kHz audio
- you want a pre-trained deep learning pitch tracker without training your own model
- you need both f0 values and voicing confidence per frame
- you prefer a simple CLI or Python API for audio analysis

## When to avoid
- you need polyphonic pitch estimation for multi-instrument audio
- you cannot install TensorFlow or lack GPU/CPU resources for deep inference
- you need real-time low-latency pitch tracking in production
- you need a lightweight non-neural pitch tracker

## Facets
- artifact type: library
- maturity: maintenance
- function: machine-learning, audio-processing, cli
- domain: machine-learning
- platform: python, cli, cross-platform
- tags: pitch-estimation, monophonic-pitch-tracking, tensorflow, keras, pretrained-model, music-information-retrieval, fundamental-frequency, audio

## Member repositories
- marl/crepe (main) score 23

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:04:38.943085+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T04:38:24.839129+00:00, confidence not recorded.
  - readme: https://github.com/marl/crepe (fetched 2026-08-28T04:04:38.943085+00:00, sha e7866c83da6d)
  - homepage: https://marl.github.io/crepe/ (fetched 2026-08-29T11:51:43.074580+00:00, sha 18fb7403d80a)
  - registry_pypi: https://pypi.org/pypi/crepe/json (fetched 2026-08-29T11:51:43.083695+00:00, sha 5088e9f64b69)
- Data as of 2026-08-30T08:39:29.467469+00:00.