Dataset Preview
Duplicate
The full dataset viewer is not available (click to read why). Only showing a preview of the rows.
Job manager crashed while running this job (missing heartbeats).
Error code:   JobManagerCrashedError

Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.

audio
string
duration
float64
semi-label
string
corrected-label
string
text_maya
string
"{'bytes': b'RIFF\\xd2)\\x01\\x00WAVEfmt \\x10\\x00\\x00\\x00\\x01\\x00\\x01\\x00D\\xac\\x00\\x00\\x(...TRUNCATED)
0.864
I
<BRUITS>
<BRUITS>
"{'bytes': b'RIFFF\\x8c\\x07\\x00WAVEfmt \\x10\\x00\\x00\\x00\\x01\\x00\\x01\\x00D\\xac\\x00\\x00\\x(...TRUNCATED)
5.608
"Kumaw yɛrɛ kɛ cogo fɛ e n'a dɔn ko jigi t tɛe la ni'i yɔrɔ'i la ko in nakɔ ye ne fa i t t(...TRUNCATED)
"Kumaw yɛrɛ kɛcogo fɛ e n'a dɔn ko jigi t'e la niyɔrɔ t'i la ko in na koyi ne fa i ti diɲɛ (...TRUNCATED)
"Kumaw yɛrɛ kɛcogo fɛ e n'a dɔn ko jigi t'e la niyɔrɔ t'i la ko in na koyi ne fa i ti diɲɛ (...TRUNCATED)
"{'bytes': b'RIFF\\x16\\xb8\\x03\\x00WAVEfmt \\x10\\x00\\x00\\x00\\x01\\x00\\x01\\x00D\\xac\\x00\\x0(...TRUNCATED)
2.763
E de y'an denw bila tula taga la
E de y'an denw bila tugan la tagala.
E de y'an denw bila tugan la tagala.
"{'bytes': b'RIFF\\x9c\\x05\\x03\\x00WAVEfmt \\x10\\x00\\x00\\x00\\x01\\x00\\x01\\x00D\\xac\\x00\\x0(...TRUNCATED)
2.245
Aw jamana yɛrɛ kɔnɔna na ala ye nin
Hli jamana yɛrɛ kɔnɔna na a lajɛ kɛ.
Hli jamana yɛrɛ kɔnɔna na a lajɛ kɛ.
"{'bytes': b'RIFF*\\x01\\x01\\x00WAVEfmt \\x10\\x00\\x00\\x00\\x01\\x00\\x01\\x00D\\xac\\x00\\x00\\x(...TRUNCATED)
0.746
Taa'aw ɲini
Taa'w ɲini
Taa'w ɲini
"{'bytes': b'RIFFT\\xdd\\x00\\x00WAVEfmt \\x10\\x00\\x00\\x00\\x01\\x00\\x01\\x00D\\xac\\x00\\x00\\x(...TRUNCATED)
0.642
Don cɛ
__ Ou bien __ cɛ...
__ Ou bien __ cɛ...
"{'bytes': b'RIFF\\x96%\\x08\\x00WAVEfmt \\x10\\x00\\x00\\x00\\x01\\x00\\x01\\x00D\\xac\\x00\\x00\\x(...TRUNCATED)
6.053
Ko'fa filɛ se waati ko y'a yati a fa ten
<MUSIQUE>
<MUSIQUE>
"{'bytes': b'RIFF\\xa8D\\x03\\x00WAVEfmt \\x10\\x00\\x00\\x00\\x01\\x00\\x01\\x00D\\xac\\x00\\x00\\x(...TRUNCATED)
2.428
Nin tuloloma bɛɛ araj sa efmu kɔrɔ bi
Nin tuloma be __ Radio Sahel FM __ kɔrɔ bi
Nin tuloma be __ Radio Sahel FM __ kɔrɔ bi
"{'bytes': b'RIFF\\x92d\\r\\x00WAVEfmt \\x10\\x00\\x00\\x00\\x01\\x00\\x01\\x00D\\xac\\x00\\x00\\x88(...TRUNCATED)
9.951
"Awze bilainic tanrazesinra manarain an sinsintala tɔgɔ barikakɛma ye masaba fɔlɔna de b'a laba(...TRUNCATED)
"<INVOCATION> an b'an sinsin ni Alahutala tɔgɔ barikama ye masa min b'a fɔlɔ n'ale b'a laban an (...TRUNCATED)
"<INVOCATION> an b'an sinsin ni Alahutala tɔgɔ barikama ye masa min b'a fɔlɔ n'ale b'a laban an (...TRUNCATED)
"{'bytes': b'RIFF\\xb8\\x93\\x11\\x00WAVEfmt \\x10\\x00\\x00\\x00\\x01\\x00\\x01\\x00D\\xac\\x00\\x0(...TRUNCATED)
13.06
"Ɲɔgɔn faga ma ka y'a kɔnɔ cogo si la an musow yɛrɛ dama k'an ka ɲɔgɔn tɛgɛ minɛ ka ɲ(...TRUNCATED)
"Ɲɔgɔn faga man ka y'a kɔnɔ cogoya si la, anw musow yɛrɛdama k'an ka ɲɔgɔn tɛgɛ minɛ ka(...TRUNCATED)
"Ɲɔgɔn faga man ka y'a kɔnɔ cogoya si la, anw musow yɛrɛdama k'an ka ɲɔgɔn tɛgɛ minɛ ka(...TRUNCATED)
End of preview.

Kunkado Maya V2 (Standardized)

Description

Ce dataset est une version transformée du dataset original RobotsMali/Kunkado (configuration human-reviewed). L'objectif de cette version est de fournir des transcriptions dont les tags d'émotions et de bruits sont 100% compatibles avec les standards d'intégration de Maya One.

Transformations effectuées

Une pipeline de nettoyage automatique a été appliquée sur la colonne corrected-label pour générer la colonne text_maya :

  1. Standardisation des tags : Passage du format Kunkado [TAG] ou <TAG> vers le format Maya <tag>.
  2. Mapping des émotions :
    • COLERE / PEUR $\rightarrow$ <cri>
    • JOIE / RIRE $\rightarrow$ <laugh> (ou <rire>)
    • APPEL $\rightarrow$ <invocation>
    • INCANTATION $\rightarrow$ <incantation>
    • SILENCE $\rightarrow$ <silence>
  3. Nettoyage du bruit : Suppression totale des tags de type incompréhensible.
  4. Normalisation : Suppression des espaces multiples et mise en minuscules des tags.

Structure du Dataset

Le fichier contient les colonnes originales pour référence, ainsi que la nouvelle colonne cible :

  • text_maya : La transcription finale prête pour Maya One.

Utilisation

from datasets import load_dataset
dataset = load_dataset("binaryMao/Kunkado_Maya_V2")
Downloads last month
29