Canary Song Decoder: Transduction and Implicit Segmentation with ESNs and LTSMs - Archive ouverte HAL Access content directly
Conference Papers Year : 2021

Canary Song Decoder: Transduction and Implicit Segmentation with ESNs and LTSMs

(1) , (1)
1
Nathan Trouvain
  • Function : Author
Xavier Hinaut

Abstract

Domestic canaries produce complex vocal patterns embed- ded in various levels of abstraction. Studying such temporal organization is of particular relevance to understand how animal brains represent and process vocal inputs such as language. However, this requires a large amount of annotated data. We propose a fast and easy-to-train trans- ducer model based on RNN architectures to automate parts of the anno- tation process. This is similar to a speech recognition task. We demon- strate that RNN architectures can be efficiently applied on spectral fea- tures (MFCC) to annotate songs at time frame level and at phrase level. We achieved around 95% accuracy at frame level on particularly complex canary songs, and ESNs achieved around 5% of word error rate (WER) at phrase level. Moreover, we are able to build this model using only around 13 to 20 minutes of annotated songs. Training time takes only 35 seconds using 2 hours and 40 minutes of data for the ESN, allowing to quickly run experiments without the need of powerful hardware.
Fichier principal
Vignette du fichier
TrouvainHinaut2021_ICANN_Canary-decoder_HAL-v2.pdf (457.04 Ko) Télécharger le fichier
Origin : Files produced by the author(s)

Dates and versions

hal-03203374 , version 1 (20-04-2021)
hal-03203374 , version 2 (23-12-2021)

Identifiers

Cite

Nathan Trouvain, Xavier Hinaut. Canary Song Decoder: Transduction and Implicit Segmentation with ESNs and LTSMs. ICANN 2021 - 30th International Conference on Artificial Neural Networks, Sep 2021, Bratislava, Slovakia. pp.71--82, ⟨10.1007/978-3-030-86383-8_6⟩. ⟨hal-03203374v2⟩

Collections

CNRS INRIA INRIA2
166 View
285 Download

Altmetric

Share

Gmail Facebook Twitter LinkedIn More