Podcast Episodes
Back to Search
Why Kitchen Smells Hijack ADHD Focus Like Noise
Episode 6545
Adults with ADHD often smell just as well as anyone else, yet brain imaging reveals their olfaction-associated regions handle those signals different…
2 days, 7 hours ago
Inside Laptop Microphones and Why They Hurt Dictation
Episode 6544
Almost every mainstream laptop—from the ThinkPad E15 to the Dell XPS 13 and Surface Laptop 7—ships with just two microphones, leaving only the MacBoo…
2 days, 7 hours ago
When Fine-Tuning Over-Generalizes Instead of Forgetting
Episode 6543
Fine-tuning a speech model on IP addresses can make ordinary phrases like "ten one" transcribe as "10.1." Rather than catastrophic forgetting, this q…
2 days, 7 hours ago
Five Projects to Build a Language Model From Scratch
Episode 6542
Training a language model from scratch used to require massive budgets, but a thousand-fold drop in compute costs has changed the math. A de facto fi…
2 days, 8 hours ago
Donating Bandwidth Without Becoming an Exit Node
Episode 6541
Donating bandwidth to help people bypass internet censorship doesn't mean turning your home connection into a risky exit node. Tools like Tor's Snowf…
2 days, 8 hours ago
Serving Parakeet Speech Models on Cheap CPUs
Episode 6540
Deploying custom speech models does not require keeping an expensive GPU running around the clock. Exploring NVIDIA's Parakeet TDT 0.6B architecture …
3 days, 4 hours ago
When Real Engineering Teams Still Choose Encoders
Episode 6539
While decoder models dominate headlines, production engineering teams frequently rely on bidirectional encoders for classification, tagging, and retr…
3 days, 6 hours ago
How Models Score Next Tokens All at Once
Episode 6538
Scoring every word in a vocabulary seems computationally ruinous, but language models and speech recognizers compute every token's probability simult…
3 days, 6 hours ago
Expecting Gidi, Act Three: Twenty-Eight Days
Episode 6537
An audio drama returns to the bench outside a shuttered café to track the real-life aftermath of Israel's election as of November 11. With the presid…
3 days, 23 hours ago
Why Speech Models Need More than Fifty Hours
Episode 6536
Folk wisdom suggests automated speech recognition fine-tuning stops paying off after fifty hours of audio, but model architecture tells a different s…
4 days ago