Analyst finds Ohio Lindner call likely authentic, not AI-generated
kevinnbass · x · 2026-10-12
kevinnbass published a forensic analysis of the three released "Lindner call" recordings tied to Talarico in Ohio, concluding the audio is likely a genuine phone call rather than AI-generated speech.
Method: signal measurements, two model listening passes, and four open-source deepfake-speech classifiers with positive/negative controls.
Key evidence:
- Dual-channel signatures: the staffer's voice is hard-cut at 3.8–4.2 kHz (narrowband phone network) while the caller extends to 8–10 kHz (wideband mic), consistently across every turn — a fake would need to synthesize two voices and two channel models coherently;
- Continuous line noise: the telephone-band noise floor shows no steps through a 6-second pause and the key sentence, where a splice would typically leave discontinuities;
- Single source: the two short clips are re-encodings of the 6:40 file's tail (sample correlation 0.986/0.998) with no internal splices;
- Call log match: the screenshot's calls to Lindner's published number match the recording's length and IVR prompt.
The author notes this only answers "is it AI-generated" and partially "did the call happen" — whether the content is accurate cannot be tested by audio.
More from Models
- Burkov: OpenAI's Luna seems deliberately weak as a multi-step agent — burkov · 2026-10-12
- Rumors claim Google, OpenAI and Anthropic are all near recursive self-improvement — FinanceYF5 · 2026-10-12
- Fine-tuning Gemma 3 E2B with LoRA on a 16GB gaming rig in 12 minutes — tomaarsen · 2026-10-12
- Perplexity's decider model adds Nimble acknowledgment after credit dispute — antoine_chaffin · 2026-10-12
- Meta unveils Muse Glimmer, a 30B open agentic model built for on-device use — emmanuelvivier · 2026-10-12
- Cloudflare open-sources Clef decision models, tops Jev Decision Index — emmanuelvivier · 2026-10-12