Another amazing week in Voice AI

Voice AI weekly digest

Top Updates 💪

• Salesforce & ServiceNow invest $1.5bn in Genesys (CX Today)

• Apple’s new Siri may allow users to operate apps just using voice (TechCrunch)

• NVIDIA releases open dataset, models for multilingual speech AI (NVIDIA Blog)

• Voice AI is changing how we do business, starting with loans (Forbes)

• Cerence Audio AI and HARMAN AudioworX boost in-car experiences (Nasdaq)

• Pocket FM gives writers an AI tool to transform narratives & more (TechCrunch)

• Aqua Voice shows just how good Mac dictation could be (9to5Mac)

• Deepgram receives 2025 contact center technology award (Morningstar)

• Krisp launches Audio-only, small turn-taking model for Voice AI Agents (Krisp)

• NiCE and Salesforce deepen end‑to‑end customer service partnership (No Jitter)

• The Opus Research conversational AI / self‑service Intelliview 2025 (CX Today)

• Singular Hearing launches HeardThat Plus to fill transcript gap (PR Newswire)

• Palabra gets backing from Reddit co-founder’s venture firm (TechCrunch)

• AI interpreting startup Kotoba raises $11.8M in in seed (Slator)

• Maven Voice brings human-like AI to support calls (PR Newswire)

• Capacity acquires Call Criteria and Verbio Technologies (Smart Customer Service)

• LG adds Cerence AI to smart TVs (TV Technology)

• Nvidia aims to solve AI issues with many languages (Artificial Intelligence News)

• Voice AI is becoming the streaming industry’s secret weapon (Streaming Media)

• Google partners with projects to bridge Asian language barriers (Newsbytes.PH)

Our Latest Article

---

Engineering Corner 😎

• AI scans audio recordings to detect voice box cancer (The Scientist)

• Whisfusion: Parallel ASR decoding via a diffusion transformer (arXiv)

• SpeakerLM: End-to-end versatile speaker diarization and recognition with multimodal large language models (arXiv)

• NTT’s 18 papers accepted for Interspeech2025, the world’s largest international conference on spoken language processing (WebWire)

• FFmpeg 8.0 merges OpenAI Whisper filter for automatic speech recognition (Phoronix)

• Building IVR systems with Vodia’s JavaScript IVR (Telecom Reseller)

• A multimodal affective interaction architecture integrating BERT‑based semantic understanding and VITS‑based emotional speech synthesis (MDPI)

• Pitch accent detection improves pretrained ASR (arXiv)

• Real‑time accessibility platform with Redis‑powered intelligence (DEV)

• Snowflake AI_TRANSCRIBE: Transform audio to insights with SQL (DEV)

• AVE Speech: A comprehensive multimodal dataset for speech recognition integrating audio, visual, and electromyographic signals (IEEE Xplore)

• Building a production-ready STT system with fine-tuned Whisper (DEV)

• Voice Embed transforms text to embeddable voice players (Trend Hunter)

• Dubbing movies via hierarchical phoneme modeling and acoustic diffusion denoising (ResearchGate)