Photos, Video and Audio: Research Tools and Source Verification

DOJ Epstein Library and disclosure files

The official library links disclosure datasets, court records and House releases. It warns that search can be unreliable for some formats and that released material may still contain sensitive information.

1. Choose DOJ Disclosures and the relevant dataset or court collection.
2. Match the complete filename and EFTA identifier from your lead. Check whether the result is a PDF representation or a playable native file.
3. Retain the original URL and page or timestamp. Preserve official redactions in any research notes or publication.

Example: For a media lead labeled EFTA00179446, use the archive’s source link and confirm the same identifier on the DOJ destination rather than assuming a screenshot is the source.

Read the result: A release filename is a reference, not authentication of every statement in the recording. DOJ says audio redactions can be a steady tone; do not mistake that tone for natural silence or an editing discovery.

Access: Public government files have no purchase charge. Age confirmation and content warnings apply. DOJ disclosure collections.

Epstein Data Image Analysis

The image catalog describes extracted photographs and scanned material using automated fields for objects, settings, visible text and other content. It can help locate an object when an EFTA number is unknown.

1. Open Image Analysis or the Images filter in document search.
2. Search a concrete object or setting, such as “painting” or “staircase”, rather than an inferred accusation.
3. Open the candidate record, note its EFTA number and page, then examine the actual source image.

Example: Try “painting” and compare a candidate’s catalog description with the linked page. A description is a search aid, not an art attribution.

Read the result: AI descriptions can misread objects, text and context. Different image totals on landing pages and tables can reflect different processing stages. Do not turn a search count into a count of events or people.

Access: Public landing page with no payment or login prompt observed. Some table links currently redirect to the project repository; use public document search as a fallback.

This guide credits Tommy Carstensen’s archive, R. Howard Stone’s Epstein Data project, and the Jmail contributors for their finding aids. Jmail credits Matheus Mendes for JeffTube. Tool makers’ descriptions and community annotations should be checked against the original media before a factual claim is published.

Tommy Carstensen’s Image Gallery

The gallery combines extracted page scans and photographs from the DOJ datasets. It offers filters and file identifier lookup, with links back to document pages and official PDFs.

1. Enter the complete EFTA identifier in Search File ID.
2. Narrow by dataset or PDF page count if useful, keeping a record of the active filters.
3. Open an image and follow View Document Page and the source PDF to recover its surrounding context.

Example: Look up “EFTA00056410” to explore images associated with that document rather than treating each extracted image as a separate source file.

Read the result: Scanned pages are not necessarily photographs. Automated person labels are expressly uncertain, and this guide does not use them for identification. Gallery badges also need their underlying explanation checked.

Access: Public gallery opened without a login or payment prompt. Respect censored tiles and the archive’s removal policy.

This guide credits Tommy Carstensen’s archive, R. Howard Stone’s Epstein Data project, and the Jmail contributors for their finding aids. Jmail credits Matheus Mendes for JeffTube. Tool makers’ descriptions and community annotations should be checked against the original media before a factual claim is published.

Tommy Carstensen’s Video and Audio Galleries

Separate audio and video galleries provide identifier search, filters and available machine transcripts. The video page explains that some originals were re-encoded for web playback.

1. Search the EFTA identifier or narrow by dataset, duration and format.
2. Open the recording and any available transcript. Read the source description and follow the DOJ link.
3. Listen to the passage and surrounding context. Save a timestamp and identify whether you used an original or playback conversion.

Example: Video: “EFTA00179446”. Audio: “EFTA00028983”, listed as a short news recording. These are file lookups, not topic assertions.

Read the result: Whisper transcripts are finding aids. A playable copy may use a different codec from the original. Silence, clipped speech and poor recording quality can defeat transcription; a timestamp needs checking in the version cited.

Access: Public galleries opened without a login or payment prompt. Not every recording has speech or a usable transcript.

This guide credits Tommy Carstensen’s archive, R. Howard Stone’s Epstein Data project, and the Jmail contributors for their finding aids. Jmail credits Matheus Mendes for JeffTube. Tool makers’ descriptions and community annotations should be checked against the original media before a factual claim is published.

Tommy Carstensen’s Video and Audio Galleries

Separate audio and video galleries provide identifier search, filters and available machine transcripts. The video page explains that some originals were re-encoded for web playback.

1. Search the EFTA identifier or narrow by dataset, duration and format.
2. Open the recording and any available transcript. Read the source description and follow the DOJ link.
3. Listen to the passage and surrounding context. Save a timestamp and identify whether you used an original or playback conversion.

Example: Video: “EFTA00179446”. Audio: “EFTA00028983”, listed as a short news recording. These are file lookups, not topic assertions.

Read the result: Whisper transcripts are finding aids. A playable copy may use a different codec from the original. Silence, clipped speech and poor recording quality can defeat transcription; a timestamp needs checking in the version cited.

Access: Public galleries opened without a login or payment prompt. Not every recording has speech or a usable transcript.

This guide credits Tommy Carstensen’s archive, R. Howard Stone’s Epstein Data project, and the Jmail contributors for their finding aids. Jmail credits Matheus Mendes for JeffTube. Tool makers’ descriptions and community annotations should be checked against the original media before a factual claim is published.

TinEye

TinEye searches for copies and altered versions of an image on the web. It is useful for finding an older publication, a larger copy or a crop that lost its original caption.

1. Upload an appropriate public image or submit its image URL.
2. Compare Best Match with Oldest, Biggest Image and Most Changed results.
3. Open the source pages and check the original caption, credited photographer and publication context.

Example: Search a non-sensitive crop of a publicly circulated property photograph. Compare the full image with the crop before making any location claim.

Read the result: “Oldest” is the earliest date TinEye crawled a matching image, not the date it was created or first published. No match means no match in its index. Search results do not grant reuse rights.

Access: TinEye states that non-commercial use is free; commercial use uses its paid API. Access terms and result sorting.

House Oversight video and transcript releases

The committee’s release pages provide the provenance of its own published deposition videos and transcripts. This is the preferred starting point for a congressional session.

1. Find the official release using the witness and session date.
2. Open the full video or transcript linked by the committee. Distinguish deposition date from publication date.
3. Check a clip against the full exchange, keeping the question, answer, timestamp and transcript page together.

Example: The March 2, 2026 release identifies the Hillary Clinton and Bill Clinton depositions as February 26 and February 27 respectively. Use those dates to distinguish the sessions.

Read the result: A committee press release, a question asked by a member and a witness’s answer are different evidence. A transcript does not establish that every statement is true.

Access: Public committee pages and linked public materials have no purchase fee. A video player may require JavaScript. Example official transcript release.

Epstein Data Reverse Image Search

The Image / Scene Search mode compares visual content with indexed released material. Use it to trace a room, object or already public non-sensitive image across the corpus.

1. Select Image / Scene Search. Do not use Face Search for this workflow.
2. Upload or paste an appropriate scene image in a supported format, then run the search.
3. Inspect candidate matches and follow their source identifiers. Compare the whole image, visible features and surrounding pages.

Example: Use a crop containing only a publicly released painting or architectural feature to look for related views of that object or room.

Read the result: The similarity score measures resemblance, not the probability that a caption is true. Similar furniture or layouts can create false matches. A match does not prove location, date, ownership or authenticity.

Access: Public tool page, with JPEG, PNG and WebP uploads up to 25 MB. The service says uploads are processed on its server; it logs request metadata. Read the on-page privacy notice before using it.

This guide credits Tommy Carstensen’s archive, R. Howard Stone’s Epstein Data project, and the Jmail contributors for their finding aids. Jmail credits Matheus Mendes for JeffTube. Tool makers’ descriptions and community annotations should be checked against the original media before a factual claim is published.

Epstein Data audio, transcript and deposition tools

The transcript collection lists full transcripts and time aligned segments. The separate Congressional Depositions tool synchronizes searchable transcripts with video and explicitly warns that speaker identification is automated and approximate.

1. Search a known EFTA identifier in document search, or open the relevant deposition.
2. Search a distinctive phrase, then move to the matching segment or video position.
3. Listen before quoting. For congressional testimony, compare the passage with the official committee transcript when available.

Example: Use “EFTA00179446” to locate a known media file. In a deposition, search a short phrase from an official transcript rather than a broad surname.

Read the result: Searchable words and speaker labels are derived data. False starts, accents, noise and redaction tones can produce errors. Table links may redirect; the search interface and linked source recording remain the checks that matter.

Access: Public landing pages opened without a paid account. Some media and synchronized controls require JavaScript; not all recordings are transcribed.

This guide credits Tommy Carstensen’s archive, R. Howard Stone’s Epstein Data project, and the Jmail contributors for their finding aids. Jmail credits Matheus Mendes for JeffTube. Tool makers’ descriptions and community annotations should be checked against the original media before a factual claim is published.

Epstein Data audio, transcript and deposition tools

The transcript collection lists full transcripts and time aligned segments. The separate Congressional Depositions tool synchronizes searchable transcripts with video and explicitly warns that speaker identification is automated and approximate.

1. Search a known EFTA identifier in document search, or open the relevant deposition.
2. Search a distinctive phrase, then move to the matching segment or video position.
3. Listen before quoting. For congressional testimony, compare the passage with the official committee transcript when available.

Example: Use “EFTA00179446” to locate a known media file. In a deposition, search a short phrase from an official transcript rather than a broad surname.

Read the result: Searchable words and speaker labels are derived data. False starts, accents, noise and redaction tones can produce errors. Table links may redirect; the search interface and linked source recording remain the checks that matter.

Access: Public landing pages opened without a paid account. Some media and synchronized controls require JavaScript; not all recordings are transcribed.

This guide credits Tommy Carstensen’s archive, R. Howard Stone’s Epstein Data project, and the Jmail contributors for their finding aids. Jmail credits Matheus Mendes for JeffTube. Tool makers’ descriptions and community annotations should be checked against the original media before a factual claim is published.

JTube / JeffTube and JPhotos

Jmail’s official About page calls its video app JeffTube. JPhotos is its photo browsing companion. These interfaces reorganize released material into familiar media app layouts.

1. Open the relevant app and choose a file or public release collection.
2. Record the displayed original identifier. On a JeffTube watch page, follow View original on justice.gov.
3. Compare the official file and its context before citing a caption, clip or image.

Example: The EFTA01683329 watch page identifies the original EFTA number and a DOJ source link. Do not infer content from the filename alone.

Read the result: App view counts, comments and the “Jeffrey Epstein” channel styling are features of the archive interface, not proof of original publication or authorship. The main app uses JavaScript and can initially show only a loading screen.

Access: Public pages were reachable without a paid account; full playback and app controls depend on browser support. Official suite description and credits.

This guide credits Tommy Carstensen’s archive, R. Howard Stone’s Epstein Data project, and the Jmail contributors for their finding aids. Jmail credits Matheus Mendes for JeffTube. Tool makers’ descriptions and community annotations should be checked against the original media before a factual claim is published.

JTube / JeffTube and JPhotos

Jmail’s official About page calls its video app JeffTube. JPhotos is its photo browsing companion. These interfaces reorganize released material into familiar media app layouts.

1. Open the relevant app and choose a file or public release collection.
2. Record the displayed original identifier. On a JeffTube watch page, follow View original on justice.gov.
3. Compare the official file and its context before citing a caption, clip or image.

Example: The EFTA01683329 watch page identifies the original EFTA number and a DOJ source link. Do not infer content from the filename alone.

Read the result: App view counts, comments and the “Jeffrey Epstein” channel styling are features of the archive interface, not proof of original publication or authorship. The main app uses JavaScript and can initially show only a loading screen.

Access: Public pages were reachable without a paid account; full playback and app controls depend on browser support. Official suite description and credits.

This guide credits Tommy Carstensen’s archive, R. Howard Stone’s Epstein Data project, and the Jmail contributors for their finding aids. Jmail credits Matheus Mendes for JeffTube. Tool makers’ descriptions and community annotations should be checked against the original media before a factual claim is published.

Jotify and the official Maxwell interview collection

Jotify presents an audio browsing interface with release and proceeding playlists. When using the Maxwell interview material, compare it with the official DOJ interview collection, which supplies transcripts and audio parts.

1. Choose the relevant playlist or official interview day.
2. Identify the recording part and the matching transcript, keeping day and sequence together.
3. Listen to the complete answer and surrounding question before recording a quotation.

Example: Open the Maxwell Proffer Sessions playlist, then match the day and part with the corresponding DOJ release. Keep it separate from Maxwell trial testimony.

Read the result: Playlist titles are organizational labels. An interview, proffer and courtroom testimony are different proceedings. A claim made in a recording requires its own corroboration.

Access: The public playlist page and DOJ collection opened without a payment prompt. This workflow does not alter or replace EpsteinWiki’s separate Proffer Index.

This guide credits Tommy Carstensen’s archive, R. Howard Stone’s Epstein Data project, and the Jmail contributors for their finding aids. Jmail credits Matheus Mendes for JeffTube. Tool makers’ descriptions and community annotations should be checked against the original media before a factual claim is published.

Jotify and the official Maxwell interview collection

Jotify presents an audio browsing interface with release and proceeding playlists. When using the Maxwell interview material, compare it with the official DOJ interview collection, which supplies transcripts and audio parts.

1. Choose the relevant playlist or official interview day.
2. Identify the recording part and the matching transcript, keeping day and sequence together.
3. Listen to the complete answer and surrounding question before recording a quotation.

Example: Open the Maxwell Proffer Sessions playlist, then match the day and part with the corresponding DOJ release. Keep it separate from Maxwell trial testimony.

Read the result: Playlist titles are organizational labels. An interview, proffer and courtroom testimony are different proceedings. A claim made in a recording requires its own corroboration.

Access: The public playlist page and DOJ collection opened without a payment prompt. This workflow does not alter or replace EpsteinWiki’s separate Proffer Index.

Google Lens

Google Lens can surface websites containing an image or similar visual content. It complements corpus search when the question is where a non-sensitive scene or object appeared publicly.

1. At Google, choose Search by image and upload a suitable image or paste its image URL.
2. Inspect the matching websites and compare the entire source image, not just thumbnails.
3. Use a specific text refinement for a visible object or location lead, then verify that lead against independent records.

Example: Search a crop containing only a public building facade or artwork. Add a neutral term such as “townhouse” if the results are too broad.

Read the result: Visually similar images and AI summaries are leads, not authentication. Uploads disclose the image to the provider; do not submit private photographs, survivor identities or intimate material.

Access: Google documents a public browser workflow without a paid subscription step. Features and account prompts can vary. The official instructions explain upload and URL search.