Gnani AI launched Timbre v3.0 on October 14, 2026, adding inline expression tags to multilingual text-to-speech for businesses and creators. The model combines 13 production voices with streaming audio, contextual number reading, and beta support for switching languages within a sentence.
Gnani AI launched Timbre v3.0 on October 14, 2026, according to Analytics India Magazine's reporting. The text-to-speech model supports 10 languages, including Hindi, Bengali, Gujarati, Kannada, Malayalam, Marathi, Punjabi, Tamil, Telugu, and Indian-accented English. Gnani's product page lists 13 production voices and positions the release for both creator narration and enterprise voice agents.
Developers can insert audio tags directly into a script to control how individual lines sound, including laughter, sighs, whispers, shouts, pauses, and changes in pace. The model can also generate non-verbal sounds such as coughs, sniffles, yawns, and exhales. For text without explicit tags, Analytics India Magazine reports that the model adjusts delivery using the sentence's meaning, tone, and punctuation, with tags supported across its languages.
Timbre v3.0 includes text normalization for information that voice agents must read accurately, such as card numbers, policy identifiers, addresses, dates, currencies, and financial abbreviations. Its product page also describes beta code-switching between an Indian language and English within a sentence while preserving the voice's character. These capabilities target multilingual calls where a natural delivery must coexist with clear pronunciation of customer and transaction details.
Gnani offers REST, streaming, and real-time WebSocket interfaces, with playback able to begin before the complete response has been generated. The company advertises first audio in under 150 milliseconds on its product page, while the launch reporting describes latency under 200 milliseconds. Alongside customer support, payment reminders, and appointment calls, the release supports long-form narration, audiobooks, podcasts, voiceovers, and tools that read digital content aloud.
