
Te encuentra — available as voice only and full mix.
- Recording
- Te encuentra
- Origin
- synthetic
- Language
- es-ES
- Notes
- Synthetic provenance confirmed by the owner on 5 November 2026.

Synthetic recordings
Listen to the voice. Hear the difference.

JOIA — Voice Studio

Te encuentra — available as voice only and full mix.

Just go — available as voice only and full mix.

Thunder — available as voice only and full mix.

Viene contigo — available as voice only and full mix.
Loading audio…
One vocal identity. New ways to produce
Speech synthesis developed from JOIA’s original recordings.
JOIA Voice Studio is an independent production environment for selected projects that need a consistent Spanish vocal identity, script flexibility and alternative takes.
Each production is guided by a creative brief and the capabilities of the selected model. Voice only and Full mix are two listening versions of the same piece.
Listen to the workExplore alternative readings, refine pronunciation and update approved scripts without recording every version from scratch.
V3 provides a conversational baseline. V4 explores reference-conditioned emotional expression experimentally; capabilities vary by model and are reviewed for each production.
Synthetic voice production, alternative versions and authorised voice datasets. Scope and licensing are agreed for every project.
01 / Selected synthetic productions
A familiar voice. A different way to produce
JOIA’s own vocal identity, developed through synthesis for selected Spanish-language productions. A flexible workflow for consistent content, script updates and alternative deliveries.
02 / Production & versions
Revisit a passage. Explore another take
Alternative takes, script updates and production versions within the validated capabilities of the selected model and agreed use.
03 / Authorised voice data projects
The recording matters. So does what comes after
Professional Spanish voice material for speech technology, model development and research. Recording, file preparation and metadata organised around your technical requirements and agreed usage.
Built from a human voice. Developed to explore what comes next
JOIA Voice Studio began with a simple question: how much of a performance can a synthetic voice preserve?
The project has grown into an independent speech synthesis environment, built around recordings of my own voice and developed through successive model iterations.
It combines curated voice data, expressive fine-tuning, pronunciation control and a custom audio production pipeline.
The goal isn’t simply to generate speech. It’s to explore how timing, character and intention carry through the process.
Independent voice technologyQwen3-TTSMLX / LoRACustom DSP
A custom Castilian Spanish voice built on Qwen3-TTS around JOIA’s own vocal identity. Recognisable timbre and speaking characteristics give approved productions a consistent voice, from a single script to recurring content.
Four experimental iterations, V1–V4, explore different ways to deliver a script. V3 provides a conversational baseline; V4 explores reference-conditioned emotional expression, offering further creative options within each model’s expressive range.
Shape pace, pauses and pronunciation around the intention of your script. Configurable generation parameters and phonetic corrections support alternative takes and multilingual passages, helping refine the delivery rather than settling for the first result.
Curated voice data and custom emotional datasets designed for the model place vocal character and expressive variation at the centre of development. Local adaptation through LoRA and MLX on Apple Silicon supports an independent, custom voice workflow.
A local FastAPI and React production environment organises scripts, takes and reusable generation configurations by project. Independent block generation lets you revisit a passage without regenerating the whole script, making longer productions and alternative versions easier to manage.
Custom DSP and mastering bring loudness normalization, dynamic processing and true-peak control into one delivery pipeline. Configured targets of −14 LUFS integrated and −1 dBTP, with WAV and MP3 export, support consistent audio preparation across approved productions.
Tell me about your script, intended use and production needs.