StreamingMemeStreamingMeme
LeaderboardsEventsSubmit News
SUBSCRIBE

Daily Brief

The streaming industry in your inbox every morning.

Daily Brief

The streaming industry in your inbox every morning.

StreamingMeme

The streaming technology industry news aggregator.

About UsNewsletterSubmit News
© 2026 StreamingMeme. All rights reserved.
← AI for Video
AI & VideoTechnical DevelopmentMay 17, 2026

Whisper runs locally on Apple Silicon with no network access

Whisper runs locally on Apple Silicon with no network access
ayushchat

OpenAI's Whisper speech-to-text model can run entirely on-device on Apple Silicon, leveraging the Neural Engine and Unified Memory for real-time transcription without network access. This local implementation maintains model accuracy while offering benefits like zero latency, data privacy, and no per-minute cost compared to the cloud API. The article details the Whisper pipeline, model sizes, and performance trade-offs on different Apple chips, noting M2 devices can transcribe 10 minutes of audio in approximately 63 seconds.

Key Takeaways

  • Whisper is described as an encoder-decoder transformer trained on 5 million hours of audio.
  • On Apple Silicon, the full pipeline runs locally: mic audio, mel spectrogram, encoder, decoder, and output text.
  • Model sizes range from Tiny at 39M parameters and about 75 MB to Large-v3 at 1.55B parameters and about 2.9 GB of RAM.
  • For M2 devices, the article says 10 minutes of audio can be transcribed in about 63 seconds.
  • The OpenAI Whisper API costs $0.006 per minute, while the local version has zero per-minute cost and zero data transmission.

Why It Matters

This shows speech-to-text can move from cloud calls to fully local execution on Macs without changing the underlying Whisper model. For teams shipping dictation, captioning, or transcription features, the trade-off is now mostly between RAM, speed, and chip class rather than model access itself. The article also notes that some cloud dictation products post-process Whisper output through an LLM, which can rewrite non-English text; on-device use returns raw output. What to watch: how M1, M2, M3, and M4 performance compares in real workloads, especially the model size each chip can sustain.


Read full article at reddit.com

Related Articles

Broadcast: AMD pushes AI to the edge for live broadcast latency and trust
Startuphub: Wasmer builds Node.js edge runtime in two weeks using OpenAI Codex
Spotify Engineering: Spotify: 99% of Engineers Use AI Coding Tools Weekly, Productivity Up 76%

Newest

in about 1 hour
The Broadcast Bridge: Decoding H.264: Navigating AVC Profiles, Levels, and Signaling for Streaming
about 10 hours ago
Valkey: Momento CTO Details Valkey's Role in High-Scale Streaming and AI Caching
about 10 hours ago
Cloudinary: Cloudinary Publishes Guide for Migrating Media Assets to Its Platform
about 10 hours ago
Azure Player: AzurePlayer Updates Focus on Video Playback Optimization for Streaming Professionals
about 10 hours ago
Upwork: Upwork Spotlights GLSL Specialists for Video Processing and Edge AI
about 10 hours ago
F6s: Global-M Platform Monetizes Video Content for Over 25 Operators
about 10 hours ago
Broadcastbeat: Mediaproxy LogServer Adds Nielsen CBET for Radio Audience Measurement
about 10 hours ago
Senza Fili: Wi-Fi Alliance Certifies 6 GHz Wi-Fi 6E, Boosting Streaming Throughput and Latency
about 10 hours ago
Span: XFRA to Deploy 1 Gigawatt of AI Compute in Homes by 2027
about 10 hours ago
Intelmarketresearch: Edge Computing Market to Exceed $31B by 2034, Driven by 5G and Immersive Media
about 10 hours ago
Fierce Network:
about 10 hours ago
SQLServerCentral: Edge AI is for Constraints, Not for Aesthetics, New Report Warns
about 10 hours ago
YouTube: YouTube Shorts Launches AI 'Dream Screen' for Background Generation
about 10 hours ago
Network World: Edge Computing, Private 5G Cut Live Event Latency to Under 5ms
about 10 hours ago
YouTube: Canada Reverses 15% Streaming Revenue Requirement Amid U.S. Trade Pressure
about 10 hours ago
YouTube: Canada Backs Off CRTC's Triple Fee Hike on Streamers Amid US Trade Concerns
about 10 hours ago
In: Amazon Plans Indiana Data Center Expansion, Faces Environmental Review
about 10 hours ago
BeBee: Hearst's WTAE-TV Seeks Digital Sales Manager to Drive AI-Enhanced Revenue
about 11 hours ago
LinkedInEditors: NETINT: VPU performance hinges on system architecture, not just silicon efficiency
about 11 hours ago
Microsoft: Microsoft Foundry Unveils MAI-Voice-2 AI for Multilingual Speech Generation

Upcoming Events

Jun
8–11
NEM Dubrovnikhttps://neweumarket.com/dubrovnik/
Jun
11–12
Arctic 15https://arctic15.com/
Jun
13–19
InfoCommhttps://www.infocommshow.org/
Jun
16–19
Stream TV Show (formerly the Pay TV Show)https://www.streamtvshow.com/
Jun
17–19
Content Tokyo 2024https://www.content-tokyo.jp/ja-jp.html
View all events →

Top Sources

  1. 1.MSN155
  2. 2.Calendly106
  3. 3.Sports Video Group65
  4. 4.Advanced Television63
  5. 5.TV Technology42
  6. 6.Cord Cutters News41
  7. 7.Broadband TV News36
  8. 8.AOL34
Full leaderboards →

Newest

in about 1 hour
The Broadcast Bridge: Decoding H.264: Navigating AVC Profiles, Levels, and Signaling for Streaming
about 10 hours ago
Valkey: Momento CTO Details Valkey's Role in High-Scale Streaming and AI Caching
about 10 hours ago
Cloudinary: Cloudinary Publishes Guide for Migrating Media Assets to Its Platform
about 10 hours ago
Azure Player: AzurePlayer Updates Focus on Video Playback Optimization for Streaming Professionals
about 10 hours ago
Upwork: Upwork Spotlights GLSL Specialists for Video Processing and Edge AI
about 10 hours ago
F6s: Global-M Platform Monetizes Video Content for Over 25 Operators
about 10 hours ago
Broadcastbeat: Mediaproxy LogServer Adds Nielsen CBET for Radio Audience Measurement
about 10 hours ago
Senza Fili: Wi-Fi Alliance Certifies 6 GHz Wi-Fi 6E, Boosting Streaming Throughput and Latency
about 10 hours ago
Span: XFRA to Deploy 1 Gigawatt of AI Compute in Homes by 2027
about 10 hours ago
Intelmarketresearch: Edge Computing Market to Exceed $31B by 2034, Driven by 5G and Immersive Media
about 10 hours ago
Fierce Network:
about 10 hours ago
SQLServerCentral: Edge AI is for Constraints, Not for Aesthetics, New Report Warns
about 10 hours ago
YouTube: YouTube Shorts Launches AI 'Dream Screen' for Background Generation
about 10 hours ago
Network World: Edge Computing, Private 5G Cut Live Event Latency to Under 5ms
about 10 hours ago
YouTube: Canada Reverses 15% Streaming Revenue Requirement Amid U.S. Trade Pressure
about 10 hours ago
YouTube: Canada Backs Off CRTC's Triple Fee Hike on Streamers Amid US Trade Concerns
about 10 hours ago
In: Amazon Plans Indiana Data Center Expansion, Faces Environmental Review
about 10 hours ago
BeBee: Hearst's WTAE-TV Seeks Digital Sales Manager to Drive AI-Enhanced Revenue
about 11 hours ago
LinkedInEditors: NETINT: VPU performance hinges on system architecture, not just silicon efficiency
about 11 hours ago
Microsoft: Microsoft Foundry Unveils MAI-Voice-2 AI for Multilingual Speech Generation

Upcoming Events

Jun
8–11
NEM Dubrovnikhttps://neweumarket.com/dubrovnik/
Jun
11–12
Arctic 15https://arctic15.com/
Jun
13–19
InfoCommhttps://www.infocommshow.org/
Jun
16–19
Stream TV Show (formerly the Pay TV Show)https://www.streamtvshow.com/
Jun
17–19
Content Tokyo 2024https://www.content-tokyo.jp/ja-jp.html
View all events →

Top Sources

  1. 1.MSN155
  2. 2.Calendly106
  3. 3.Sports Video Group65
  4. 4.Advanced Television63
  5. 5.TV Technology42
  6. 6.Cord Cutters News41
  7. 7.Broadband TV News36
  8. 8.AOL34
Full leaderboards →