Dear Jonathan,
TTS can indeed be a good option. We have compared synthetically generated speech material with the original material (using the
MOS scale). This was also done for Spanish, and we found similar scores. Note that these systems generally have a limited bandwidth (up to 8 kHz).
Polspoel, S., Moore, D. R., Swanepoel, D. W., Kramer, S. E., & Smits, C. (2025).
Automatic development of speech-in-noise hearing tests using machine learning. Scientific
Reports, 15(1), 12878.
Best,
Cas Smits
Van: AUDITORY - Research in Auditory Perception <AUDITORY@xxxxxxxxxxxxxxx>
Namens Massimo Grassi
Verzonden: zaterdag 7 februari 2026 09:23
Aan: AUDITORY@xxxxxxxxxxxxxxx
Onderwerp: Re: [AUDITORY] Spanish Speech Material (Male)
Dear Jonathan,
why you do not consider the text-to-speech AI? They seem very good to my ears. Actually, I wanted to use these sounds in an experiment in which I want to present speech in various languages with the same voice-timbre. In other words, (and
this is also a question I address to the list members, possibly experts in this topic) are these new AI voices bad?
All the best to everybody and greetings from the alps, very close to the olympics.
Dear list members,
We are currently preparing a study investigating speech-intelligibility scores in listeners with normal hearing.
The study will be run in Spain and will use the Castilian Spanish HINT material as target speech (Huarte, 2008).
We are looking for speech recordings in Castilian Spanish by male speakers, to be used as competing-talker maskers.
We are for example looking for recordings of male speakers reading news stories, books, etc...
Any resource you may be aware of would be highly appreciated!
Reference: Huarte, A. (2008). “The Castilian Spanish Hearing in Noise Test,” Int J Audiol,
47, 369–370. doi:10.1080/14992020801908269
______________________________________________________
AmsterdamUMC disclaimer : www.amsterdamumc.org/nl/disclaimers.htm
|