Text-to-speech
Also written: TTS
Software that reads digital text aloud, converting written words into spoken audio.
Text-to-speech, or TTS, is a technology that takes written words and turns them into spoken audio. Think of it like a digital narrator that can read any text you give it out loud.
This is useful for many things, like making digital content more accessible for people who have trouble reading or for listening to articles while you do other tasks. For example, you could use a TTS tool to have a news article read to you on your phone.
Text-to-speech (TTS) systems convert written text into synthesized speech. Modern TTS models, often based on deep learning architectures like transformers or diffusion models, can generate highly natural-sounding human speech. These systems typically involve stages like text normalization, phonemization, and acoustic modeling to produce the final audio waveform.
Tradeoffs include latency, computational cost, and the ability to control prosody (intonation, rhythm, stress). While many systems offer basic voice options, advanced models can be trained on specific voices for personalization, raising ethical considerations around deepfakes.
Recent advancements in AI models like Google's Gemini 2.5 Pro and OpenAI's GPT-4o Mini highlight improved text-to-speech capabilities. These developments focus on delivering more natural and studio-quality speech synthesis, making the technology more accessible and useful for various applications.
- Gemini 2.5 Pro TTSGoogle · Gemini Models ·
- GPT-4o Mini TTS ModelOpenAI · Model Documentation ·
- SchoolAI builds an AI platform that empowers teachersOpenAI · OpenAI News ·
- Expanding on how Voice Engine works and our safety researchOpenAI · OpenAI News ·
- Google Gemini
- Google Gemini's 2.5 Pro TTS offers studio-quality speech synthesis for professional audio needs.
- OpenAI ChatGPT
- OpenAI's GPT-4o Mini TTS model converts text into natural speech, and their Voice Engine is a custom voice model focused on safety.
- Speech-to-text (STT)
- This is the opposite process, converting spoken audio into
- Voice assistants
- These are AI systems that can understand and respond to spoken
The concept of converting text to speech has roots in early computing and assistive technology research dating back to the mid-20th century. Early systems were often rule-based and robotic, with significant advancements driven by machine learning and deep learning in recent decades.
Open a document or webpage on your phone or computer. Most operating systems have a built-in TTS feature you can enable in accessibility settings. Once enabled, select some text and use the feature to have it read aloud to you.
The Playbook: reusable how-tos, prompts, configs and tips pulled out of the news.
- ToolGemini 3.8 Flash TTS for custom voice narrationUse Gemini 3.8 Flash TTS via the Gemini API in Google AI Studio for custom voice narration and vocal affectations. · 15 min
- Code snippetGemini 3.8 TTS API Schema UpdateUpdate API requests for Gemini 3.8 TTS models from earlier versions. · 5 min
- Code snippetUse custom voice in xAI Text to Speech APIPass the custom voice_id as the 'voice' query parameter in TTS API calls. · 3 min
- Troubleshooting fixUpdate Gemini TTS API Schema for Gemini 3.8Migrate from older Gemini TTS models to Gemini 3.8 by updating your API request schema. · 10 min
- Cost tipCost-effective audio generation with Gemini TTSGenerate longer audio clips efficiently using Gemini 3.8 TTS models. · 5 min
Questions people ask
- Is TTS the same as a voice assistant?
- No, TTS is a component that reads text aloud. A voice assistant uses TTS to speak its responses, but also uses speech-to-text to understand your commands.
- Can TTS sound like a real person?
- Modern TTS systems can sound very natural and human-like, often indistinguishable from a real person in short samples.
First seen on Subvolts Jun 7, 2024. Updated Oct 19, 2026. Written by an AI model from the items above and reviewed against them weekly; report an error.