Simba 3.5

Simba 3.5 is an expressive text-to-speech model developed by SpeechifyAI. Released on October 21, 2026, it serves as a major expansion within the Simba model family by unifying support for 48 languages under a single model identifier. This capability allows any voice in the Speechify catalog—as well as custom voice clones—to fluently render speech across all supported regional variants and languages.

Unlike previous iterations that limited specific voices to designated languages, Simba 3.5 ensures that every catalog voice maintains its native accent and quality while speaking globally. The model supports real-time streaming speech, word-level timing data, and SSML configurations for precise pitch and volume adjustments. Developers can integrate the model via the SpeechifyAI API using the simba-3.5 model identifier for tasks requiring high-throughput, low-latency audio generation.

Integration into applications is managed through endpoints supporting various audio formats such as MP3, OGG, AAC, and PCM. Simba 3.5 is designed to maintain backward compatibility with previous versions regarding voice IDs and parameter schemas, making it straightforward for developers to transition existing workflows to the expanded multilingual framework.