WhisperDirect vs SecureTalkiOS

WhisperDirect is an audio transcription and summarization app built around the high accuracy of the Whisper API. In addition to the Whisper API using your own API key, it includes on-device features such as Apple Speech, Speaker Diarization (NVIDIA Nemotron 3), and OCR. For summaries and meeting minutes, choose from OpenAI / Gemini / OpenAI-compatible LLMs, using APIs only as much as you need. One-time purchase — no subscription. Pricing & Experience ・7-day free trial included ・On-device features (Apple Speech / Speaker Diarization / OCR) are free, no API key required ・Whisper API and LLM (Summary / Minutes) are billed by each provider on a usage basis 　(The app does not charge for API usage) Cost Estimates ・Whisper API: ~$0.006/min (1 hour ≈ $0.36) ・Segment re-transcription (gpt-transcribe): ~$0.0045/min ・On-device features incur no API fees ・Actual costs vary with audio length and settings ・OpenAI API pricing: https://openai.com/api/pricing ・Gemini / custom providers follow each service's pricing ・Locally-run custom providers (e.g. Ollama) incur no API fees Transcription Engines ・Whisper API (OpenAI): high accuracy with your own API key ・Apple Speech (On-device): no API key, iOS 26+ ・Switchable from settings Models for Summary & Minutes ・Choose from OpenAI / Gemini / Custom 　- OpenAI: GPT-6 Luna / GPT-5 nano / GPT-5 mini / GPT-4.1 nano / GPT-4.1 mini 　- Gemini: requires Google AI Studio API key (gemini-*-pro / flash / gemma) 　- Custom: OpenAI-compatible APIs (Ollama / LM Studio / self-hosted) ・Processing costs only a few cents even for several-thousand-character text Key Features ・Record with the mic button and transcribe on the spot ・Live Preview: real-time transcription while recording (on-device) ・Recording display mode switch (Waveform / Simple) ・Import audio files → transcribe (Share Sheet) ・Video import (audio extracted and compressed) ・Transcript auto-highlights with playback position ・Insert timeline markers at any interval (5-sec steps) ・Generate summaries from transcripts ・OCR from images (batch supported, no API fees) ・Generate meeting minutes (prompts editable) ・Re-transcribe misrecognized segments in the editor 　(Uses OpenAI gpt-transcribe; billed by usage) ・Re-transcribe specific items with Whisper API, even while using Apple Speech ・Speaker diarization (NVIDIA Nemotron 3, on-device) ・Export: Audio / Text / Summary / Minutes / VTT / SRT ・Auto-post to Slack ・Webhook integration (Discord, Slack, n8n, etc.) ・Google Drive auto-backup ・Estimated cost display ・Fine-grained settings (LLM model, timeline interval, prompts) Supported Formats Audio: mp3, m4a, aac, wav, flac, ogg, opus, wma, amr, mpga, webm, aiff, caf Video: mp4, mov, m4v, webm, mkv, avi, mpeg, mpg Notes ・Whisper API requires an OpenAI API key ・On-device features work without an API key ・Prices, specs, and models may change per provider

Brand Name SecureTalk Brand Vision To ensure every conversation is fully understood in a secure, real-time, and cloud-independent environment. Brand Mission Built on an Edge AI architecture, SecureTalk provides a real-time voice AI solution that runs on standard PCs and laptops without requiring a GPU or cloud upload. It is designed to address the following challenges: ・Confidential meetings cannot be uploaded to the cloud ・Long meetings require excessive time and effort to organize ・Multilingual communication remains inefficient ・Enterprise IT security policies often prevent the adoption of SaaS-based voice services