My Weird Prompts
by Daniel Rosehill
Latest episode
How Streetcar Suburbs Hit the Transit Sweet Spot
How Streetcar Suburbs Hit the Transit Sweet Spot
About this podcast
No question too obscure, no rabbit hole too deep. A man, a sloth, and a donkey talk through one question at a time (with a lot of help from AI). My Weird Prompts celebrates curiosity in all its forms. Daniel, the human, asks the questions that pop into his head at inconvenient moments, and now joins the hosts in the studio to ask them himself. Corn the Sloth offers laid-back, thoughtful takes. Herman the Donkey brings boundless enthusiasm and energy. Together, they explore topics ranging from the mundane to the mind-bending. Every episode is made by an AI pipeline that researches the question, writes the script, voices it with AI text-to-speech and assembles the final podcast. Every voice you hear is synthetic, and some are voice clones, Daniel's included. Listeners can send prompts too. Stay curious.
Episodes
The newest 50 of 3,000 episodes
- How Streetcar Suburbs Hit the Transit Sweet Spot · 27 min · Episode 6550Streetcar suburbs were built around electric tram lines in the late nineteenth century, yet most still function decades after the rails were torn out. Their secret is a specific…
- Plugging a Custom Voice Model into Wayland · 26 min · Episode 6549Most open-source voice projects assume you need speech recognition, leaving a developer who already trained their own models with nowhere to plug them in. Exploring tools like…
- Why Home Voice Agents Still Need Home Assistant · 21 min · Episode 6548Home Assistant already speaks MCP in both directions, so demoting it to one tool behind a server-hosted voice agent is technically possible with frameworks like LiveKit Agents…
- USB Far-Field Mics and the Echo Cancellation Catch · 21 min · Episode 6547A far-field microphone array is really a small computer with microphones attached, and the DSP chip behind them is the actual product. For a DIY Home Assistant and Whisper…
- The Wake-Word Race Inside Home Assistant Satellites · 28 min · Episode 6546Multi-room Home Assistant voice setups rely on cheap satellites—like the Voice Preview Edition or ESP32-S3-BOX-3—streaming audio to one central hub. Because wake-word arbitration…
- Why Kitchen Smells Hijack ADHD Focus Like Noise · 25 min · Episode 6545Adults with ADHD often smell just as well as anyone else, yet brain imaging reveals their olfaction-associated regions handle those signals differently. This episode examines…
- Inside Laptop Microphones and Why They Hurt Dictation · 20 min · Episode 6544Almost every mainstream laptop—from the ThinkPad E15 to the Dell XPS 13 and Surface Laptop 7—ships with just two microphones, leaving only the MacBook Pro 14 with three. This…
- When Fine-Tuning Over-Generalizes Instead of Forgetting · 26 min · Episode 6543Fine-tuning a speech model on IP addresses can make ordinary phrases like "ten one" transcribe as "10.1." Rather than catastrophic forgetting, this quieter failure is…
- Five Projects to Build a Language Model From Scratch · 28 min · Episode 6542Training 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 five-step ladder—centered on…
- Donating Bandwidth Without Becoming an Exit Node · 22 min · Episode 6541Donating bandwidth to help people bypass internet censorship doesn't mean turning your home connection into a risky exit node. Tools like Tor's Snowflake, Psiphon's Conduit…
- Serving Parakeet Speech Models on Cheap CPUs · 22 min · Episode 6540Deploying custom speech models does not require keeping an expensive GPU running around the clock. Exploring NVIDIA's Parakeet TDT 0.6B architecture reveals why quantized INT8…
- When Real Engineering Teams Still Choose Encoders · 28 min · Episode 6539While decoder models dominate headlines, production engineering teams frequently rely on bidirectional encoders for classification, tagging, and retrieval. A close look at real…
- How Models Score Next Tokens All at Once · 26 min · Episode 6538Scoring every word in a vocabulary seems computationally ruinous, but language models and speech recognizers compute every token's probability simultaneously in a single matrix…
- Expecting Gidi, Act Three: Twenty-Eight Days · 29 min · Episode 6537An 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 presidential mandate officially…
- Why Speech Models Need More than Fifty Hours · 21 min · Episode 6536Folk wisdom suggests automated speech recognition fine-tuning stops paying off after fifty hours of audio, but model architecture tells a different story. The fifty-hour mark…
- Why AI Software Creation Threatens the SaaS Model · 24 min · Episode 6535Software stocks recently shed two trillion dollars over expectations that agentic AI will break traditional per-seat business models. At the same time, tools like Microsoft…
- Why People Pay to Drink Terrible Spirits · 22 min · Episode 6534Fermentation takes human ingenuity to unusual places, from industrial oddities to ancient living traditions. Across the globe, drinks like Chicago's famously bitter Malört thrive…
- How Economics Broke and Rebuilt Israeli Wine · 25 min · Episode 6533Israeli viticulture stretches back to biblical antiquity, but modern winemaking was built by Baron Edmond de Rothschild and the Carmel cooperative in the late nineteenth century…
- Train Your Wine Palate with Side-by-Side Tasting · 20 min · Episode 6532Developing an intuitive palate for wine does not require sommelier certification, complex vocabulary, or expensive cellar management tools. The most effective approach relies on…
- How Drip Irrigation Rebuilt Israeli Wine · 23 min · Episode 6531Israeli viticulture transformed from nineteenth-century sacramental bottles into a landscape of over three hundred boutique wineries. This guide tours key producers across…
- Why Hospital Wayfinding Relies on Environmental Design · 32 min · Episode 6530Poor navigation design carries hidden human and financial costs across complex spaces like healthcare facilities. Examining environmental wayfinding—from the colored floor lines…
- How Typesetting Shifted From Metal to Software · 25 min · Episode 6529For over five centuries, typesetting was a specialized manual craft defined by physical lead sorts, meticulous redistribution, and later, towering line-casting machines like the…
- Why Typography Science Keeps Disproving Type Rules · 28 min · Episode 6528Every word read in public spaces or on screens was drawn by a type designer, yet legibility science and typographic craft remain at an impasse. A look into how typefaces are…
- Why Prescribed Stimulant Withdrawal Stays Unstudied · 27 min · Episode 6527Immediate-release stimulants clear the body within hours, but the neurochemical adaptations built around daily dosing can linger for weeks. Despite clear label warnings for…
- Why wav2vec 2.0 Still Dominates Speech Downloads · 26 min · Episode 6526Modern speech leaderboards are dominated by giant models like NVIDIA Parakeet and Whisper, yet Meta's wav2vec 2.0 base checkpoint still commands roughly 1.8 million downloads a…
- Why Speech Recognition Models Struggle With Accents · 23 min · Episode 6525Speech-to-text models still benchmark against standard American and British English, treating other accents as margins of error. Longitudinal audits show newer systems often…
- Why Israel Never Finished Its Sovereign Hebrew AI · 28 min · Episode 6524Israel spent years, four national plans, and hundreds of millions of shekels trying to build a sovereign Hebrew and Arabic language model. Tracing the initiative from the TELEM…
- Unpacking Israel's National AI Strategy · 26 min · Episode 6523A close look at Israel's Resolution 4255 and its newly formed National AI Directorate reveals a massive gap between public ambition and state funding. Rather than training costly…
- Nebius and the GPU Cloud Pivot · 28 min · Episode 6522After selling its Russian business at a steep discount in 2024, Yandex's Dutch parent entity rebranded as Nebius Group N.V. Backed by billions in Nvidia capital and relocated…
- Multitenancy in Qdrant Vector Databases · 29 min · Episode 6521Multitenancy in vector databases involves four distinct isolation trade-offs: storage, compute, enforcement, and operations. When partitioning in Qdrant, the dominant cost is…
- Mapping Multi-Tenancy Across Vector Databases · 28 min · Episode 6520Approximate nearest neighbor indexes scan neighborhoods without knowing who owns a vector, forcing vector databases to resolve multi-tenancy either inside physical structures or…
- How Smuggled GPUs Move Through Real Companies · 22 min · Episode 6519Export controls bar advanced AI chips from China, but illicit networks bypass restrictions using real, operational businesses across Southeast Asia rather than empty shell…
- Why Export Bans Cannot Stop Remote Compute · 23 min · Episode 6518United States export regulations govern the physical shipment of AI chips across borders, but they leave remote server rentals almost entirely unrestricted. In regions like…
- Why GPU Wiring Dictates Model Parallelism · 26 min · Episode 6517Training massive AI models requires keeping GPUs constantly busy, but communication bottlenecks create steep performance cliffs. Bandwidth plummets by a factor of eighteen the…
- Why Cargo Thieves Are Targeting GPU Shipments · 25 min · Episode 6516Cargo theft is shifting from opportunistic warehouse break-ins to sophisticated deceptive pickups targeting high-value computing hardware. This look into an $11.4 million Nvidia…
- Creative Schemes Behind the Global GPU Hunt · 19 min · Episode 6515Tech companies, venture capital funds, and independent researchers resort to wild gambits to secure elusive computing power amid severe chip constraints. Approaches span from…
- Why Even Solo Developers Face a GPU Shortage · 23 min · Episode 6514On-demand GPU rental capacity has quietly dried up, catching solo developers fine-tuning on a single L4 just as severely as hyperscalers. An analysis of surging memory costs…
- Why Dual-Socket Servers Win on Dense AI Models · 26 min · Episode 6513Adding a second CPU socket offers extra memory channels and PCIe lanes, but non-uniform memory access dramatically shapes real-world performance. A detailed look at Intel Xeon…
- Build Servers Never Actually Build Anything · 22 min · Episode 6512Continuous integration terminology is full of misnomers, starting with the fact that build servers rarely compile code. Evaluating Woodpecker CI, Forgejo Runner, and self-hosted…
- Why Voiceprints Are Not an Interoperable File Format · 23 min · Episode 6511Voice biometrics often promises portable voiceprints, yet no vendor-neutral standard exists to exchange speaker models. The official ISO/IEC 19794-13 biometrics standard…
- Voice Isolation Before Speech Recognition · 21 min · Episode 6510Voice isolation models strip out competing speakers before speech reaches voice activity detection and speech-to-text engines. While models like VoxLock, Scicom, Hush, and…
- How a Knesset Government Can Form with Under 61 Votes · 27 min · Episode 6509Coalition formation after the election rests on parliamentary mathematics rather than pure majorities. An incoming Israeli government does not actually require sixty-one…
- Why Speech Transducers Omit Common Symbols · 25 min · Episode 6508When speech models fail to produce common characters like email at-signs, the limitation stems from deep architectural tradeoffs rather than missing training data. RNN-T and TDT…
- Why Broadcast Headsets Fit Commentators but Not Voiceover · 21 min · Episode 6507Closed-cup broadcast headsets dominate sports commentary press boxes by locking mic distance and blocking crowd noise, but using one for home voiceover brings distinct…
- Why Generative Audio Restoration Hallucinates Highs · 28 min · Episode 6506Modern audio restoration tools like NVIDIA's RE-USE and Tokyo's Sidon ditch traditional filtering to synthesize clean speech representations from scratch. While they effortlessly…
- Why Blanket Forts Beat Expensive Portable Vocal Booths · 23 min · Episode 6505Portable vocal setups promise studio sound anywhere, but most home recording purchases disappoint because buyers confuse acoustic treatment with isolation. Blankets…
- Why Voiceover Studios Obsess Over Repeatability · 25 min · Episode 6504Professional voiceover setups are engineered around physical consistency, ensuring recordings made months apart match seamlessly. This deep dive covers microphone positioning…
- Why Tokenizer Vocabulary Size Is a Hardware Choice · 25 min · Episode 6503Before an AI model reads a word, a pre-trained tokenizer chops the text into pieces using byte pair encoding. While vocabulary size is often treated as an arbitrary convention…
- What a Cross-Entropy Loss of 1.8 Actually Means · 25 min · Episode 6502Training runs live and die by cross-entropy loss curves, yet few can translate a single loss number into plain terms. This breakdown walks through how a loss of 1.8 nats…
- Splitting Frontier AI Across Thousands of GPUs · 31 min · Episode 6501Training frontier AI models is fundamentally a distributed plumbing challenge across thousands of accelerators. When weights, optimizer states, and activations vastly exceed…
