Learning a female utterance with male and child vocal tracts

Top left: "yayayaya" spoken by a female German speaker
Top right: Synthetic "yayayaya" generated by VocalTractLab with articulatory parameters of a male speaker, learned directly from the acoustics (MFCC) of the female utterance, with no speaker normalization
Upper middle left: Synthetic "yayayaya" generated by VocalTractLab with articulatory parameters of a child, learned directly from the acoustics (MFCC) of the female utterance, with no speaker normalization
Upper middle right: Synthetic "yayayaya" generated by VocalTractLab with articulatory parameters of an even younger child, learned directly from the acoustics (MFCC) of the female utterance, with no speaker normalization
Bottom: Comparison of natural (EMA) and synthetic (VocalTractLab) trajectories of tongue tip, tongue mid and tongue back in "yayayaya". Black: original; Red: synthetic

Original, Adult German female
Synthetic, Adult male vocal tract
Spectrogram
Sound

Synthetic, Child vocal tract 1
Synthetic, Child vocal tract 2
Spectrogram
Sound

Articulatory trajectories synthesized from learned articulatory targets (red) in comparison with natural trajectories (black)