Cohere launches an open source voice model specifically for transcription

✨TodayOnAI’s Daily Drop

• Cohere launches an open source voice model specifically for transcription

• Mistral releases a new open source model for speech generation

• Spotify tests new tool to stop AI slop from being attributed to real artists

• 💬 Let’s Fix This Prompt

• 🧰 Today’s AI Toolbox Pick

Cohere

Cohere launches an open source voice model specifically for transcription

🚀 TodayOnAI Insight: Cohere has launched Transcribe, its first open-source speech recognition model—pairing strong accuracy with lightweight performance, signaling a push toward accessible, self-hosted enterprise AI.

🔍 Key Takeaways:

• New 2B-parameter ASR model optimized for consumer-grade GPUs and self-hosting

• Supports 14 languages, including English, Chinese, Arabic, and Japanese

• Tops Hugging Face Open ASR leaderboard with a 5.42 WER, outperforming rivals

• Achieves 61% human-evaluated win rate on transcription quality and usability

• Processes ~525 minutes of audio per minute; free via API and Model Vault

• Planned integration into Cohere’s North enterprise agent platform

💡 Why This Stands Out: Cohere is betting on efficiency over scale—delivering competitive accuracy without massive compute demands. This aligns with a broader shift toward deployable, cost-effective AI that enterprises can control. As voice becomes a core interface layer, will lightweight, open models outpace closed, heavyweight incumbents?

Mistral

Mistral releases a new open source model for speech generation

🚀 TodayOnAI Insight: Mistral AI has unveiled Voxtral TTS, an open-source text-to-speech model built for real-time, on-device voice generation—positioning itself as a low-cost, customizable alternative to OpenAI, ElevenLabs, and Deepgram.

🔍 Key Takeaways:

• Voxtral TTS supports 9 languages and can clone a voice from <5 seconds of audio, preserving accents, tone, and speech nuances

• Built on Ministral 3B, it runs efficiently on edge devices like smartphones and even smartwatches

• Delivers near-instant response with ~90ms time-to-first-audio and 6× real-time rendering speed

• Designed for enterprise use cases—sales agents, customer support, dubbing, and multilingual assistants

• Part of a broader push toward a full multimodal stack combining speech, text, and vision

💡 Why This Stands Out: Mistral is betting that open-source + edge deployment will undercut incumbents on both cost and flexibility. By enabling high-quality voice cloning in seconds and real-time performance on-device, it lowers the barrier to deploying conversational AI at scale. The bigger signal: voice is no longer a feature—it’s becoming a core interface layer for enterprise AI systems.

Spotify

Spotify tests new tool to stop AI slop from being attributed to real artists

🚀 TodayOnAI Insight: Spotify is testing a new safeguard to combat the surge of AI-generated “slop” and misattributed tracks, giving artists final approval over what appears on their profiles—an important step toward restoring identity control in streaming.

🔍 Key Takeaways:

• Spotify announced “Artist Profile Protection,” a beta feature letting artists approve or reject releases before they go live

• Only approved tracks will appear on profiles, count toward stats, and feed recommendation algorithms

• The move addresses rising issues from AI-generated music, metadata errors, and impersonation attempts

• Artists receive alerts when new music is submitted under their name via Spotify for Artists

• Comes amid industry pressure, including mass takedown requests of AI impersonation tracks

💡 Why This Stands Out:bThis signals a shift from open distribution toward controlled identity layers in streaming. As generative AI lowers the barrier to music creation, platforms are being forced to rethink trust, attribution, and ownership. The bigger question: will verification become a standard gatekeeper in the AI music era?

✨ See how a simple prompt upgrade can unlock better AI output.

🔹 The Original Prompt

At first glance, this prompt might seem okay. But it's too broad — and that limits the quality of AI-generated results. Let’s improve it using prompt engineering best practices.

✅ The Improved Prompt

💡 Why It's Better

• Specific audience: Targets decision-makers in mid-sized companies.

• Contextual focus: Emphasizes emerging tech and practical solutions.

• Actionable output: Requests summaries and titles to spark execution.

• Tone and style: Guides the type of content (insightful, engaging, relevant).

🛠️ Learn how to adapt this prompt for SaaS, AI tools, dev teams & more →
Read the full PromptPilot breakdown

💡 Bonus Tool: Want to generate and master prompts instantly?
👉 Try PromptPilot by TodayOnAI (Free to use)

📰 More from the AI World

• Mercor competitor Deccan AI raises $25M, sources experts from India

• Melania Trump wants a robot to homeschool your child

• Google launches Lyria 3 Pro music generation model

• Granola homepage Apps Granola raises $125M, hits $1.5B valuation as it expands from meeting notetaker to enterprise AI app

🧰 Today’s AI Toolbox Pick

• 🐙Learnxyz (Academics Tool): A fun, social, and causal learning app.

• 🗿Mojju (GPTs Tool): Builds specialized GPTs for productivity, creativity, and education.

• 🧘Sonia (Mental Health Tool): Provides mental health for every mind.