Server for AI audio transcription, with your private API.

It is a server prepared for AI audio transcription. It comes with Whisper, faster-whisper and WhisperX preinstalled and exposed as a private API. You send the audio and receive the text — with punctuation, timestamps and, on the GPU plan, diarization (speaker separation).

The Whisper CPU is for light volume, up to around 500 minutes of audio per day — good for transcribing meetings and one-off recordings. The Whisper GPU, with a dedicated NVIDIA GPU, processes up to around 10,000 minutes per day and does near real-time transcription, ideal for call centers, media and production pipelines.

The Whisper CPU costs US$ 65.80/mo with a one-time setup of US$ 51.80. The Whisper GPU costs US$ 519.80/mo with a one-time setup of US$ 259.80 (it covers preparing the server with the GPU, drivers and the transcription environment). No contract.

Provisioning takes up to 48 business hours. The server is prepared in a dedicated way — installing the Whisper models, the API and validation. The flow is: you request the plan, we confirm availability and delivery, and we provision it.

Yes. The audio and transcripts are processed on your server — nothing is sent to third-party APIs. That is the difference from cloud transcription services, where the audio leaves your infrastructure. Ideal for meetings, support, healthcare, legal and LGPD compliance.

Yes. Whisper was trained on multiple languages and has excellent accuracy in English, including accents and technical terms. WhisperX adds per-word timestamps and diarization — useful for subtitles and transcribing conversations with multiple speakers.

Yes. Transcription is exposed as a private REST API. You call the endpoint from within n8n workflows, your backend or any automation — to generate summaries, index content or feed other systems.

Yes — human support 24/7, with people who understand Whisper, diarization and transcription tuning. Rollin Host is a Brazilian company (Rollin Serviços Digitais e Tecnologia LTDA).