Which AI detectors detect AI voice and video in one product? The nine cards that list both, and the zero records that tested either
Nine of the 72 AI detectors in the index list both voice and video among their modalities, by the modalities field of every card read 8 and 9 November 2026 in the data file: Hive, Reality Defender, AI or Not, SynthID, TruthScan, Sightengine, Sensity, Pindrop Pulse and Resemble Detect. None of the nine has a study record on voice or on video: the three voice and video benchmarks in the index, ASVspoof 5 (arXiv 2408.08739, read 9 November 2026), SAFE 2025 (arXiv 2510.03387, read 9 November 2026) and DFDC 2020 (arXiv 2006.07397, read 9 November 2026), scored anonymised or research teams and name no indexed product, and the 4 records that do score one of the nine cards all score images: Hive 2, AI or Not 1 and Sightengine 1 (modality records ranking, read 9 November 2026).
A product that lists voice and video is answering a buyer's question about scope, and the table answers only that. Each row quotes the modalities field as the card records it from the vendor's own page, counts the records per modality from the card's study list, and links the page with its read date. Where a vendor publishes an accuracy figure for voice or video, the row's note names it as the vendor's. This page is one of the questions answered from the index; the modalities view lists every card by input type, the image and text question covers the other common pairing, and the AI detector register carries every card.
| Detector | Modalities, as the card states them | Records on voice | Records on video | Records on other modalities | Source | Read date |
|---|---|---|---|---|---|---|
| Hive | text, image, video, voice (audio), music | 0 | 0 | 2 on images: Ha and Passananti 2024, RAID image 2025 | docs.thehive.ai | 8 November 2026 |
| AI or Not | image, video, voice (audio), music, text | 0 | 0 | 1 on images: Ha and Passananti 2024, as Optic | aiornot.com | 9 November 2026 |
| Sightengine | image, video, deepfake (face swaps, separate model), voice (audio speech), music | 0 | 0 | 1 on images: RAID image 2025 | sightengine.com | 9 November 2026 |
| Reality Defender | video, image, voice (audio), text | 0 | 0 | 0 | realitydefender.com | 9 November 2026 |
| SynthID | image, video, audio, text (watermarked at generation; no public text check) | 0 | 0 | 0 | synthid.com | 9 November 2026 |
| TruthScan | text, image, PDF documents, video (deepfake), voice (audio) | 0 | 0 | 0 | truthscan.com | 9 November 2026 |
| Sensity | video, image, voice (audio), file forensics | 0 | 0 | 0 | sensity.ai | 9 November 2026 |
| Pindrop Pulse | voice (audio), video (Pulse for Meetings: "fake video and audio, synthetic avatars, and face swaps") | 0 | 0 | 0 | pindrop.com | 9 November 2026 |
| Resemble Detect | voice (audio), image, video, music (Suno and Udio listed) | 0 | 0 | 0 | resemble.ai | 9 November 2026 |
| Deepfake-O-Meter | image, video, voice (audio) | 0 | 0 | 0 | zinc.cse.buffalo.edu/ubmdfl/deep-o-meter | 10 November 2026 |
Modalities as each vendor states them on the page linked, read on the date shown; record counts from the study list on each card. The index has not uploaded a clip to any of the nine. A tenth card joined the table on 10 November 2026: Deepfake-O-Meter, the University at Buffalo research platform, runs 41 hosted detectors across image, video and audio with a login and no price, and no record scores the platform itself (Deepfake-O-Meter card, read 10 November 2026). The same wave added ElevenLabs' AI Speech Classifier, a voice-only check for the vendor's own generations, and Intel FakeCatcher, a video-only method with no public access route, both in the second table, where the eleventh wave later the same day added two more voice-only tools that listen to the audio of a playing video, Hiya Deepfake Detector (a free Chrome extension, one second of audio) and McAfee Deepfake Detector (on-device on Windows AI PCs), neither with a study record (Hiya Deepfake Detector card, McAfee Deepfake Detector card, read 10 November 2026) (ElevenLabs Classifier card, FakeCatcher card, read 10 November 2026).
Which tools detect deepfake voice and video in one product?
Nine, in three groups by what else they cover. Three are platform classifiers that also score text and images: Hive lists text, image, video, audio and music in one API (docs.thehive.ai, read 8 November 2026), AI or Not lists image, video, audio, music and text on one pricing page (aiornot.com, read 9 November 2026), and TruthScan lists text, images, PDF pages, deepfake video and audio with separate APIs for each (truthscan.com, read 9 November 2026). Four are media forensics products with no text score: Reality Defender (realitydefender.com, read 9 November 2026), Sensity, which adds file forensics on metadata and container structure (sensity.ai, read 9 November 2026), Sightengine, whose face-swap deepfake model is separate from its AI-image model (sightengine.com, read 9 November 2026), and Resemble Detect, which lists Suno and Udio among covered music generators (resemble.ai, read 9 November 2026). Pindrop Pulse began in voice and reaches video only through Pulse for Meetings, which "continuously analyzes meeting audio and video" in Microsoft Teams, Zoom and Cisco Webex (pindrop.com, read 9 November 2026). SynthID is the odd one out: it detects a watermark that Google's own models embed at generation, so its voice and video coverage extends to content from SynthID-integrated models and no further (synthid.com, read 9 November 2026).
Has any study tested these tools on voice or video?
No record in the index has, and the three benchmarks that could have did not name them. ASVspoof 5 scored 53 teams on 680,774 utterances and kept every team anonymous by policy; Pindrop appears as a co-author's affiliation and a workshop sponsor, not an entrant, and neither ASVspoof paper prints a Pindrop result (arXiv 2601.03944, read 9 November 2026; ASVspoof 5 record). SAFE 2025 ran 12 teams blind against 4,200 real and 2,600 synthetic clips and named five university teams, none of them a vendor (arXiv 2510.03387, Table 5, read 9 November 2026; SAFE 2025 record). DFDC 2020 published a leaderboard of 2,114 Kaggle teams, with no indexed vendor among them (Meta AI results post, read 9 November 2026; DFDC 2020 record). The four records that do touch one of the nine cards are image tests: Ha and Passananti 2024 recorded a 24.47% false positive rate on 280 human artworks for AI or Not under its earlier name Optic and 0.00% for Hive (DOI 10.1145/3658644.3670306, read 9 November 2026), and RAID image 2025 scored Hive and Sightengine on images (RAID image record). The leaderboard question reaches the same count from the other direction: no media card has a public entry anywhere.
What do the nine vendors claim for voice and video?
Four publish a number and five do not, and every number is the vendor's own. Pindrop states "99% (accuracy) against known engines" and "<1% false positive rate", both footnoted to "Pindrop Product Data, 2024", on its Pulse for Contact Centers page, and its audio deepfake detection page footnotes a "99% Detection accuracy against known engines" to "the ASVspoof5 dataset of 32 engines" (pindrop.com, read 9 November 2026; the card quotes both pages). Resemble Detect's benchmarks page, last updated August 2026, states "99.5% Pooled detection accuracy" on a "Podonos Audio DFD Bench", a 0.4% audio false negative rate across 4,524 files, 95.8% average on 1,000 image files and 98.2% on 1,000 video files (resemble.ai, read 9 November 2026). AI or Not's home page states "98.9% accuracy" for images, text, videos and deepfakes via API, with no split by modality (aiornot.com, read 9 November 2026). TruthScan's "99%+" figure is stated for its text detector (truthscan.com, read 9 November 2026). Hive, Reality Defender, Sensity, Sightengine and SynthID publish no accuracy rate for voice or video on the pages read; Reality Defender states that its system "grades each content with a 1-99% probability rating" and Sensity's home page calls itself "The Most Accurate Deepfake Detection Solution" without a figure. The claim vs evidence gap ranking can set none of these claims against a study, because no study exists.
Two of the nine, Reality Defender and Sensity, are compared field by field on Reality Defender vs Sensity, added 10 November 2026: 15 against 22 file types, a published price against none, cloud against on-premise forms, and no study record on either side.
Cards that list voice or video, not both
| Detector | Modalities, as the card states them | Voice | Video | Source | Read date |
|---|---|---|---|---|---|
| Proofademic | text, image (API now; browser tool coming soon), voice (API) | Yes, API | No | proofademic.ai | 8 November 2026 |
| Walter Writes | text, image (API), voice (API) | Yes, API | No | walterwrites.ai | 8 November 2026 |
| Deepware | video (face manipulation only) | No | Yes | deepware.ai | 9 November 2026 |
| Pangram | text, image, video | No | Yes | pangram.com | 8 November 2026 |
| Copyleaks | text, code, image, video | No | Yes | copyleaks.com | 8 November 2026 |
| ZeroGPT | text, image, video | No | Yes | zerogpt.com | 8 November 2026 |
| ElevenLabs Classifier | voice (audio): ElevenLabs-generated speech only | Yes, the vendor's own models | No | elevenlabs.io/ai-speech-classifier | 10 November 2026 |
| FakeCatcher | video (portrait videos, through blood-flow signals); no public access route | No | Yes | intel.com newsroom | 10 November 2026 |
| Hiya Deepfake Detector | voice (audio): the voice in a video or audio played in a Chrome tab | Yes, free extension | No, the video is the carrier | chromewebstore.google.com | 10 November 2026 |
| McAfee Deepfake Detector | voice (audio): AI-generated audio in a video playing in the browser | Yes, on-device | No, the video is the carrier | mcafee.com | 10 November 2026 |
Modalities as the card records them from the page linked, read on the date shown. The other 27 cards, read 9 November 2026 in the data file, list text, images, code or handwriting only, and none of the 6 above has a record on voice or video either.
A product that lists both voice and video has told the buyer what it accepts, and the record on what it gets right stands at zero studies for all nine.
Two siblings read the same cards from other angles: the music detectors question lists the five cards that state music, and the Pindrop vs Resemble Detect comparison sets the two voice-only vendors in the table above on the same 14 fields, both read 10 November 2026.
Questions about combined voice and video detectors
Is there one tool that detects both deepfake voice and deepfake video?
9 of 42 cards listed both on 8 and 9 November 2026, Hive, AI or Not, Sightengine, Reality Defender, SynthID, TruthScan, Sensity, Pindrop Pulse and Resemble Detect, each on the vendor page its card cites, and Deepfake-O-Meter made it 10 of 63 on 10 November 2026. SynthID's coverage is limited to content from SynthID-integrated Google models (synthid.com, read 9 November 2026).
Has any deepfake voice or video detector been tested by a study?
0 of the 9 have a record on voice or video on 9 November 2026: ASVspoof 5 (arXiv 2601.03944), SAFE 2025 (arXiv 2510.03387) and DFDC 2020 (arXiv 2006.07397) scored anonymised or research teams. The 4 records that touch Hive, AI or Not and Sightengine are image tests.
Which voice and video detector claims the highest accuracy?
99.5% pooled detection accuracy is Resemble Detect's claim for audio on a Podonos benchmark, last updated August 2026, with 98.2% for video on 1,000 files (resemble.ai, read 9 November 2026); Pindrop claims 99% against known engines, footnoted to its own 2024 product data on one page and to the ASVspoof5 dataset on another (pindrop.com, read 9 November 2026). Both figures are the vendor's own and neither appears in a study.
Does Pindrop detect deepfake video?
1 Pindrop product, Pulse for Meetings, covers video: it "continuously analyzes meeting audio and video" in Microsoft Teams, Zoom and Cisco Webex for "fake video and audio, synthetic avatars, and face swaps" (pindrop.com, read 9 November 2026); Pulse for Contact Centers and Pulse Inspect score audio.
Which detectors cover voice but not video?
2 cards, read 8 November 2026: Proofademic and Walter Writes each list voice through their API alongside text and images, with no video (proofademic.ai; walterwrites.ai). 4 cards list video and not voice: Deepware, Pangram, Copyleaks and ZeroGPT; on 10 November 2026 FakeCatcher joined them (video only) and ElevenLabs' classifier joined the voice-only side, for ElevenLabs-generated speech alone.
Sources
All read 9 November 2026 unless an earlier date is shown; the source of every row is linked in the tables.
- Hive AI-generated content detection documentation: https://docs.thehive.ai/docs/ai-generated-content-detection (read 8 November 2026)
- AI or Not pricing and home page: https://www.aiornot.com/pricing and https://www.aiornot.com/
- Sightengine AI image detection: https://sightengine.com/detect-ai-generated-images
- Reality Defender FAQ: https://www.realitydefender.com/faq
- SynthID: https://synthid.com/
- TruthScan home page and AI detector: https://truthscan.com/ and https://truthscan.com/ai-detector
- Sensity deepfake detection for video, image and audio: https://sensity.ai/deepfake-detection-for-video-image-audio/
- Pindrop Pulse for Meetings and Pulse for Contact Centers: https://www.pindrop.com/pulse-for-meetings and https://www.pindrop.com/pulse-for-contact-centers
- Resemble AI Detect and benchmarks: https://www.resemble.ai/products/detect and https://www.resemble.ai/benchmarks
- Proofademic home page: https://proofademic.ai/ (read 8 November 2026)
- Walter Writes API: https://walterwrites.ai/api/ (read 8 November 2026)
- Deepware FAQ: https://deepware.ai/us/faq/index.html
- Pangram home page: https://www.pangram.com/ (read 8 November 2026)
- Copyleaks AI Content Detector: https://copyleaks.com/ai-content-detector (read 8 November 2026)
- ZeroGPT home page: https://www.zerogpt.com/ (read 8 November 2026)
- ASVspoof 5 challenge summary and evaluation paper: https://arxiv.org/abs/2408.08739 and https://arxiv.org/abs/2601.03944
- SAFE 2025: https://arxiv.org/abs/2510.03387
- DFDC 2020 dataset paper and Meta AI results post: https://arxiv.org/abs/2006.07397 and https://ai.meta.com/blog/deepfake-detection-challenge-results-an-open-initiative-to-advance-ai/
- Ha and Passananti 2024: https://doi.org/10.1145/3658644.3670306
Revision log: 2026-11-09: page created from the modalities fields and study lists of 42 cards and the three voice and video records at index version 2026.11.8; 2026-11-10: Reality Defender vs Sensity comparison linked; 2026-11-10: Deepfake-O-Meter added to the both-modalities table, ElevenLabs Classifier (voice only) and FakeCatcher (video only) to the second table (tenth wave); 2026-11-10: Hiya Deepfake Detector and McAfee Deepfake Detector added to the voice-only table (eleventh wave); ASVspoof 2021 and ADD 2023 records added with no card scored; 2026-11-10: twelfth-wave pages linked
Related pages: Which AI detectors publish a model card or technical report? The eight cards with a dated, sized document, and the 34 with a claim instead, What input languages does Turnitin's AI detection cover? Seven, added from April 2023 to 2026, with paraphrase detection in English only, AI detector accuracy claims vs published results, ranked by the gap and Questions about AI detectors, answered from the index.