Transcription

Fathom released a significant update in May 2026 that changes the output format of its AI meeting summaries from a single flowing block of text into structured notes with clearly separated sections for context, decisions, and action items — each with assignee and deadline fields pre-populated from conversation context. The update includes automatic calendar integration: Fathom joins scheduled calls without any user interaction beyond the initial setup, meaning the tool operates as a background layer rather than requiring deliberate launch before each meeting. Action items extracted from meetings are now pushed directly to connected project management tools including Notion, Linear, and Asana, with a confidence score attached so teams can quickly identify items that need human review versus items where the extraction is unambiguous. Fathom remains free for individuals and charges teams on a per-seat basis for shared summaries, searchable archives, and the CRM integrations. For knowledge workers who spend a significant portion of their week in meetings, Fathom's May update addresses the last manual step in the transcription-to-action workflow: converting a summary into an actual task that gets done.

CraftNote shipped a feature called Speaker Memory in its May 2026 update that learns to recognize individual voices across meetings and automatically attributes every transcribed statement to the correct participant — without requiring meeting bots, calendar permissions, or manual speaker tagging each session. The system builds voice profiles locally on-device, addressing the data privacy concern that has kept compliance-sensitive organizations from using cloud-based meeting AI. Once profiles are established, CraftNote produces structured summaries that separate discussion from decisions and decisions from action items, rather than delivering a flat transcript dump that still requires manual review. The tool supports transcription in over 100 languages and works for in-person meetings without a video call, closing the gap between remote and face-to-face meeting capture. For sales teams, CraftNote also syncs action items and contact notes directly to CRM platforms, which the company says eliminates a post-meeting admin step that averages 22 minutes per meeting. The update positions CraftNote as the most privacy-forward option in a category where Otter.ai, Fathom, and Fireflies remain cloud-dependent.

Microsoft unveiled its first in-house speech-to-text model as part of a trio of MAI (Microsoft AI) foundational models released in late April 2026 — and the benchmark results are striking. The model achieves an average Word Error Rate of 3.8% on the FLEURS benchmark, the industry-standard multilingual transcription test, across the top 25 languages by Microsoft product usage. That beats OpenAI's Whisper-large-v3 on all 25 languages, Google's Gemini 3.1 Flash on 22 of 25, and ElevenLabs' Scribe v2 and OpenAI's GPT-Transcribe each on 15 of 25. For any product that needs accurate, multilingual transcription at scale — meeting notes, customer service call logs, video captions, voice-first interfaces — Microsoft now has a model it built entirely in-house that outperforms every major competitor on the industry benchmark. The strategic context matters as much as the benchmark. Until October 2025, Microsoft was contractually limited by its OpenAI partnership from independently pursuing artificial general intelligence. The MAI speech model is the clearest evidence yet that those constraints are gone and Microsoft is building seriously at the model layer. For enterprise customers running Teams, Azure, or any Microsoft communication surface, this model is expected to replace Whisper as the default transcription backend — with no additional licensing cost beyond existing Microsoft cloud contracts. Transcription tool vendors who have been building on top of Whisper should take note: Microsoft's best-in-class model is now in-house and integrated across every Microsoft product.

Eleven Labs launched Voice Cloning for customer service AI agents. Service teams can now train agents on a company's brand voice using a 1-hour recording session. Use case: Customer service chatbots sound like your company instead of generic AI. Turnaround: 60 minutes from training to deployment. Cost: $200/month per voice (includes 100 hours of generation). Early adopters (e-commerce, SaaS support): retention improved 12% when support bots match brand personality. Localization angle: Available in 27 languages; Asian markets seeing 35% adoption rate in first 2 weeks.

The new mobile feature transcribes regular phone calls (not just Zoom) with speaker ID and instant summaries. Sales teams are the target market — CRM integration included.