spINAch: A Diachronic Corpus of French Broadcast Speech Controlled for Speakers' Age and Gender

Phonetic alignments and acoustic cues of more than 3M French oral vowel estimated from the spINAch corpus 🥬 Phonetic annotations were automatically extracted by applying Whisper then Montreal Forced Aligner (MFA) on the audio files. Formants and fundamental frequency are estimated with Praat following the strategy given by Escudero et al. (2009); Data description 🔡 spinach_vowel_set.csv (aligned oral vowels) 📜 spinach_metadata.csv (speaker description) Description : spinach_vowel_set.csv : vowel : oral vowel phoneme phonetized and force-aligned by MFA with the French dictionary (resp. aligner) v3.0.0. duration : duration of the vowel ; F1 to F5 : median formant values of the middle third computed with Praat (Burg algorithm + Escudero's ceiling optimization) ; f0 : median fundamental frequency computed on the vowel's frames ; spk_id : anonymised speaker ID ; gender : gender of the speaker ; period : broadcast period of the recording ; age_5y : age of the speaker (with a five year precision) ; media : radio or tv . spinach_metadata.csv : information about the anonymised speakers like the period the speaker is interviewed, the gender and the age (within 5 years). Download instructions for the wav files WAV files and transcriptions : downloading spINAch audio files requires to accept its general term and conditions of use (GCU) and to fill this form, Citing Please cite this work if you are using these vowel acoustic cues in a publication : @Inproceedings{SPINACH_LREC26, author = "Devauchelle, Simon and Doukhan, David and Uro, Rémi and Ondel Yang, Lucas and Pelloin, Valentin and Imbert-Brégégère, Olympia and Lefort, Véronique and Picard, Kévin and Seignobos, Emeline and Rilliard, Albert", year = "2026", title = "spINAch: A Diachronic Corpus of French Broadcast Speech Controlled for Speakers' Age and Gender", booktitle={Fifteenth International Conference on Language Resources and Evaluation (LREC 2026)}, address = "Palma, Mallorca, Spain", publisher = "European Language Resources Association",}