ForgeSpy vs Pella Research

ForgeSpy checks uploaded media and social-media links for AI-generated content, deepfakes, and cloned audio. Image checks use multiple vision models and return a verdict with an AI probability score from 0–100%. Direct image uploads support JPG, PNG, WebP, and GIF. For supported social video links, the service downloads and samples frames, then provides separate visual and audio confidence scores. Its deepfake layer examines facial geometry, consistency over time, and blending artifacts to flag face swaps and identity manipulation. Audio checks look for voice cloning, text-to-speech, and AI-generated narration through spectral patterns, prosody, and acoustic artifacts. The site names Twitter/X, Instagram, YouTube, TikTok, and Facebook as supported link services. ForgeSpy is available on the web, iOS, and Android. Signing up gives five free credits without a credit card; the Free plan costs nothing, with one credit per image scan and one credit per five seconds of video. Mobile users can revisit check history and export or share reports. Detection results are estimates and should be weighed with source, context, and other verification methods.

Pella Research studies synthetic speech and builds models for detecting it in audio. Its research examines vocoder artifacts, spectral statistics, and prosodic cues. A browser demo accepts an uploaded or recorded voice clip and returns a verdict without signup; the site says audio is not stored. The REST detection API returns calibrated authenticity scores and per-segment analysis, with access reviewed after a request, typically within 24 hours. Pella offers the API to journalists, platforms, researchers, and institutions, and lists a free tier for accredited journalists and academics. Published evaluation sets track new voice-cloning systems with documented provenance and no train/test leakage. Pella says models are trained and tested with phone-line compression, background noise, and re-encoding. It reports pellav2 results of 95.82% accuracy, 0.959 F1, and 57 ms latency in the August 2026 Podonos benchmark, plus 93.6% accuracy on m4a files with errors leaning toward false positives. pellav2 is described as open weights under the MIT license.