ScriptMe

Converts audio files (MP3, WAV, M4A, OGG, FLAC, and others) into timestamped text transcripts using speech-to-text inference, likely leveraging cloud-based ASR (Automatic Speech Recognition) models or APIs. The system processes uploaded audio streams, segments them into manageable chunks, runs inference across those segments, and reassembles the output with timing metadata. This capability handles variable audio quality and sample rates through preprocessing normalization before ASR inference.

Extracts audio streams from video files (MP4, MOV, WebM, AVI, MKV) using container parsing and codec detection, then applies the same ASR pipeline as audio transcription. The system demuxes video containers to isolate audio tracks, handles variable frame rates and codecs, and optionally preserves video metadata (duration, resolution) for context. This avoids requiring users to pre-convert video to audio, reducing friction in the transcription workflow.

Provides a simple text editor interface for post-transcription corrections, allowing users to fix ASR errors, adjust punctuation, and manually add speaker labels. The editor likely operates on the transcript as plain text or simple structured data (JSON with timestamps), with changes stored back to the platform's database. No collaborative editing, version control, or advanced formatting options are mentioned, suggesting a single-user, linear editing model.

Converts transcripts from ScriptMe's internal storage format into multiple output formats (TXT, PDF, SRT, VTT, DOCX) for compatibility with downstream tools and workflows. The system likely maintains a canonical transcript representation (possibly JSON with timestamps and speaker metadata) and applies format-specific serializers to generate each output type. SRT and VTT exports include timing information for subtitle integration with video players.

Implements a quota system that tracks free-tier user consumption (transcription minutes, file uploads, storage) and enforces limits by blocking further uploads or processing when quotas are exceeded. The system likely maintains per-user counters in a database, checks quotas before accepting uploads, and displays remaining quota in the UI. Upgrade prompts are triggered when users approach or exceed limits, driving conversion to paid tiers. No transparent documentation of quota limits is mentioned, suggesting opaque tier boundaries.

Handles user file uploads (audio and video) with validation, virus scanning, and storage in a cloud backend (likely AWS S3, Google Cloud Storage, or similar). The system validates file types and sizes before acceptance, scans uploads for malware, stores files with encryption at rest, and manages retention policies (auto-deletion after processing or after a retention period). Upload progress tracking and resumable uploads may be supported for large files.

Indexes transcripts for full-text search, allowing users to find specific words, phrases, or timestamps within their transcript library. The system likely maintains an inverted index (keyword → transcript ID, timestamp) in a search engine (Elasticsearch, Solr, or database full-text search) and returns results with context snippets and playback timestamps. Search results may be ranked by relevance or recency, and filters may allow narrowing by date, speaker, or file type.