Synthetic voices specialized in Spanish and its varieties

Spanish is one of the most widely spoken languages in the world, with more than 500 million speakers across the globe. Yet most current voice technologies don’t capture the rich diversity of accents and expressions that characterize our language.

At Monoceros Labs we’re building a catalog of digital, or synthetic, voices that express the nuances of each region. The catalog is made up mainly of voices with the nuances of Spanish from Spain, followed by other voices with an Andalusian accent or that speak Mexican Spanish.

The problem: the diversity of Spanish

Spanish is characterized by its regional linguistic diversity. That diversity shows in variations in intonation and rhythm, in each region’s particular phonetics, and in local expressions and idioms. How can technology preserve those nuances and represent them?

Speech synthesis technology (TTS) has to adapt to these nuances. It’s trained on data that includes voices grouped by accent and region, which share phonetic patterns. Some specific aspects also need work, including phonetic dictionaries and text normalization, among others.

The solution: our voice catalog

At Monoceros Labs we’ve created voices specially designed to capture the richness of Spanish. These voices also have AI-generated identities and are part of our catalog of predesigned voices.

In 2024 the catalog had 14 different predesigned voices, which you can try in our online TTS editor, Fonos.

Predesigned catalog TTS voices added in 2023

Predesigned catalog TTS voices added in 2024

TTS voices available in Fonos

Voice name	Spanish variety	Pitch	Style
Altair	Spain	Low	General purpose
Betel	Spain	High	General purpose
Capella (*)	Spain	Medium	General purpose
Cástor	Spain	Medium	General purpose
Deneb	Spain	Low	General purpose
Enif	Spain	Medium	General purpose
Fulu	Spain	Medium	General purpose
Ginan	Spain	Medium	General purpose
Hatysa	Spain	Medium	Expressive, narrative
Izar	Andalusia	High	Expressive, narrative
Jabbah	Spain	Low	Expressive, narrative
Libertas	Mexico	Medium	General purpose
Maia	Spain	High	Expressive, narrative
Naos	Spain	Low	Expressive, narrative
Vega	Spain	Medium	Expressive, narrative

(*) This voice will be available on the platform soon.

Technological innovation

Proprietary technology

These synthetic voices were created with our own technology, so you won’t find them on other TTS platforms. This technology, developed under our project “Speech synthesis in Spanish for building natural conversational systems,” was funded by CDTI through its NEOTEC program.

Since 2021, our speech synthesis technology has evolved significantly in expressiveness and naturalness. It keeps pace with the latest academic state of the art (usually in English), applied to Spanish and its dialectal diversity.

1. First generation (2021-2023)

   • Design of the first AI-generated voice identities.
   • Neural synthesis models with GANs.
   • A focus on naturalness and expressiveness.
   • Prosody control.

2. Current advances (2024-2025)

   • Improved generative adversarial network (GAN) models.
   • Integration of diffusion models.
   • BERT-style language models.
   • Improvements in expressiveness and naturalness.

Accessing and using the catalog voices

Online web editor: Fonos

• Intuitive interface.
• Choose voices by style and variety.
• Intonation adjustments based on the text.

API

• Quick, simple integration.
• Programmatic control.
• Content at scale.

Use cases: who has used our voices

1. Media such as RTVE, PRISA and IDEAL (Grupo Vocento).
2. Education, such as high schools, schools and universities.
3. Companies, for marketing and content creation, such as LLYC and t2ò.

Main applications of our synthetic voices

1. Accessibility

• Adapted screen readers.
• Voice web navigation.
• Accessible educational content.
• Blog posts you can listen to.

2. Dynamic content

• Breaking news and updates.
• Conversations in voice assistants and chatbots.
• Personalized content.
• Real-time event narration.

More information

Our synthetic voice catalog is a significant step forward in voice technology for Spanish. By understanding and expressing the nuances of each region, we make technology more accessible, natural and effective for Spanish speakers around the world.

• Fonos
• Nieves Ábalos: “To reach more people we need to use conversation and voice,” in ComputerWorld (11/20/2024) (in Spanish)

The voice catalog is available through the Fonos web editor and via API, making it possible to create unique voice experiences tailored to each project’s specific needs.