Updates from Amazon, MS, Stability, Podcastle, Sesame and more! 🔥

Voice AI weekly digest

Top Updates 💪

• How Amazon combines models, agents, and browser for smarter AI (VentureBeat)

• Microsoft Dragon Copilot launches a unified voice AI for healthcare (Microsoft)

• A year later, OpenAI still hasn’t released its voice cloning tool (TechCrunch)

• Stability AI optimizes its audio model to run on ARM chips (TechCrunch)

• Podcastle launches a TTS model with more than 450 AI voices (TechCrunch)

• Talking with Sesame’s AI voice companion is amazing and creepy (ZDNet)

• How AI is changing the way we communicate – the future of interaction (Forbes)

• Deutsche Telekom and ElevenLabs announce strategic partnership to power AI-driven podcasting in Magenta App (ElevenLabs)

• This AI Meeting Assistant is a must-have for remote work (MakeUseOf)

• AI-powered chatbots revolutionizing customer experience in 2025 (SaveDelete)

• Breaking language barriers with Intercom translations (AutoGPT)

• Agora unveils AI engine for smooth voice interactions (PR Newswire)

• Deepgram Nova-3 medical AI speech model reduces healthcare transcription errors (Artificial Intelligence News)

• Fish Audio debuts Fish Speech 1.5 for realistic voice cloning (WICZ)

• Whispp lets people with speech impairments speak in real-time using AI (Heise)

• Audioshake unveils AI to separate overlapping voices (Podcasting Today)

• Uniti AI raises $4M in seed funding (Finsmes)

• HeyMilo raises $2.2M in seed funding (Finsmes)

• AI at the microphone: the voice of the future? (HIIG Digital Society Blog)

• Tightrope launches MediaScribe caption and translation service (TVNewsCheck)

Notable on X

---

Engineering Corner 😎

• Spark-TTS: an efficient LLM-based text-to-speech model with single-stream decoupled speech tokens (ArXiv)

• SpeechCompass: enhancing mobile captioning with diarization and directional guidance via multi-microphone localization (ArXiv)

• Nexus-O: An omni-perceptive and interactive model for language, audio, and vision (ArXiv)

• Enhancing dictation with conformer-based speech recognition (Medium)

• Large language model-based generative error correction (NVIDIA Research)

• How to create a voice cloning AI model for realistic speech synthesis (Dev)

• Meetily: The open-source, self-hosted alternative to FirefliesAI (Dev)

• Building a meeting summarizer backend with Python, FastAPI, AWS Transcribe, and Bedrock (Dev)

• Top AI STT tools for researchers and educators (EducatorsTechnology)