April 2023 Summaries

ElevenLabs has unveiled Eleven Multilingual v1, an advanced speech synthesis model supporting seven new languages: French, German, Hindi, Italian, Polish, Portuguese, and Spanish. This model builds on the research behind Eleven Monolingual v1, utilizing more data, computational power, and novel techniques to deliver a sophisticated and emotionally expressive performance. This advancement enables creators, game developers, and publishers to produce localized and accessible content, fostering creativity and innovation. The model, available on the Beta platform, is capable of handling multilingual text, maintaining unique voice characteristics across languages, and is compatible with features like Instant Voice Cloning and Voice Design. Despite its capabilities, it has limitations with numbers, acronyms, and foreign words defaulting to English, which the team is working to improve. ElevenLabs aims to make human-quality AI voices accessible in every language, empowering users to create high-quality audio experiences that bridge cultural gaps, enhance education, and increase accessibility for diverse audiences.

ElevenLabs has introduced a new multilingual model for Speech Synthesis that now supports seven additional languages: Spanish, French, Italian, German, Polish, Hindi, and Portuguese. This model enables the generation of speech in multiple languages using a single prompt, and Creator subscribers and above can access higher quality audio outputs at 96 kbps. The platform also features an AI SDR that efficiently qualifies 78% of inbound sales leads and is available 24/7 in over 30 languages, facilitating instant responses and meeting bookings. Additionally, ElevenLabs has demonstrated voice cloning in 12 Indian languages live at IIT Delhi, showcasing the ease and authenticity of their technology.

ElevenLabs is set to launch Professional Voice Cloning later this year, enhancing its Creator, Pro, and Scale plans by enabling users to create high-fidelity digital replicas of their voices using extensive audio data for training. This new feature promises more accurate voice reproduction compared to Instant Cloning, which operates with short samples. It is particularly beneficial for audiobook narrators, news anchors, and radio DJs, as the resulting speech is nearly indistinguishable from the original. Users can currently submit recordings for training via VoiceLab, with the Pro Voices being released to accounts sequentially starting in July. The company has also showcased its voice cloning capabilities in 12 Indian languages during a live demonstration at IIT Delhi, emphasizing the process's authenticity, ease, and speed.