12 Services from medien.halowerk.com

Performs speaker diarization on an audio or video file, returning a merged turn-by-turn transcript with speaker labels, timestamps, duration, word count, and text per turn.

Converts text to speech audio in MP3, WAV, Opus, FLAC, or raw linear16 format using a freely chosen voice from a catalog of 100+ multilingual voices, returning base64-encoded audio with its media type and duration.

Converts documents between text formats via pandoc, or extracts text/renders pages from PDFs using poppler

Reads and cryptographically verifies the C2PA manifest store of a media file, reporting manifest presence, signature validity, certificate details, and ingredient provenance chain.

Removes the background from an image (URL or base64) using rembg/U2-Net locally, returning a PNG or WebP with real alpha channel plus bounding box and crop options

Inspects all AcroForm fields of a PDF (names, types, values, options, constraints) or fills specified fields and returns the completed PDF as base64 with before/after field values.

Transcribes an audio or video file (by URL or base64) into full text with per-word timestamps, confidence scores, utterances, paragraphs, and detected language.

Analyzes a video (by URL or base64) using FFmpeg frame-difference analysis to detect scene boundaries and return timestamps, durations, and detection scores for each scene.

Translates a document (PDF, DOCX, ODT, HTML, or Markdown) from a URL, base64, or raw text into a target language while preserving all structural elements such as headings, tables, placeholders, and code blocks.

Compares 2–8 images using three perceptual hashing algorithms (average hash, gradient hash, DCT hash) to detect duplicates and near-duplicates across each pair

Compares two documents paragraph-by-paragraph, identifying added, removed, and reworded blocks with word-level change highlighting plus structural metadata counts.

Extracts hazard codes, classifications, and identifiers from Safety Data Sheets (SDS/MSDS) by downloading a PDF URL or accepting pre-extracted text, then running pdftotext to deterministically pull codes with their source lines.