The retrieval layerfor all yourspoken context.

Public interviews, private recordings, and all conversational data, transcribed, speaker-labeled, and searchable - for you and your AI agents.

How it worksSign up

Select your data.

Choose from pre-built public datasets and or upload your personal recordings - all managed from one console.

regest

Public Datasets

Choose which curated datasets to include in search.

  • All public sources18k hrs

    Include every public dataset in search.

  • UK Politics & Policy4.2k hrs

    Ministers, committees, and government briefings.

  • Tech & Business Podcasts6.1k hrs

    CEO interviews, earnings calls, and industry talks.

  • Central Banking2.8k hrs

    Fed, ECB, and BoE speeches and press conferences.

regest

Personal

Your uploaded recordings — private to your account.

  • Q1 Client InterviewsReady
    12 files
  • Board Meetings 2026Needs labels
    3 files
  • Research callsProcessing
    5 files

Upload your audio.

Add recordings to your collections, label descriptions and seed speaker names. Regest then does the rest. Building clean, context-aware, searchable indexes.

regest

Personal

Your uploaded recordings — private to your account.

  • Q1 Client InterviewsReady
    12 files
  • Board Meetings 2026Needs labels
    3 files
  • Research callsProcessing
    5 files

Connect once.

Add a single MCP config to your AI tools. Every agent can now pull source-verified quotes - public or private.

Paste into config
{
"mcpServers":{
"regest":{
"command":"npx"
"args":[
0:"-y"
1:"mcp-remote"
2:"https://mcp.regest.io/mcp"
]
}
}
}

Every quote, fully attributed.

Speaker labels, timestamps, and rich metadata on every result — helping your agents ground responses on source verified context.

API responseChristine Lagarde
{
"search_summary":{
"total_candidates":48
"top_k":1
"strategies_used":[
0:"semantic"
1:"speaker_filter"
]
"date_range":{
"min":"2024-01-01"
"max":"2025-06-01"
}
}
"results":[
0:{
"statement_text":"A retail CBDC would sit alongside cash — not replace it — and must preserve privacy by design."
"transcript_passage_text":"On digital currencies, our view is clear: a retail CBDC would sit alongside cash, not replace it, and must preserve privacy by design while meeting supervisory requirements."
"speaker":"Christine Lagarde"
"episode_title":"ECB press conference — monetary policy"
"collection_id":"public-eu-central-bank"
"release_date":"2025-03-14"
"media_type":"interview"
"source_api":"rss_api"
"audio_api":"audio_api"
"start_seconds":842
"end_seconds":867
}
]
}

Create your free personal data store now.

Sign up now

Questions

How is regest different to other tools?

Most upload-and-transcribe tools stop at a flat transcript. Regest is built for retrieval: specialised conversational diarization labels who is speaking, adverts and bumpers are stripped at ingestion, and every utterance is indexed at sentence level with rich metadata. We also build graph networks of context within your uploads — so material in one collection cross-pollinates with the rest of your corpus. Search stays precise on small sets and becomes highly intelligent as your dataset grows.

How does regest process the data?

We know that finding the statement you need, when you need it, can be hard. Regest prepares and indexes your audio at a sentence level, so search can be precise and effective at surfacing hidden statements and aha moments. Upload anything — interviews, meetings, podcasts, private recordings — once or as you receive it; our system handles the rest. We transcribe accurately, remove ads where present, and intelligently label who is speaking (although you can seed speaker names upfront and edit labels at any time). Vector representations are stored so finding a specific sentence stays fast, even across large collections.

Do you use our data?

No. Never. Your uploads and queries are yours — we do not train on your audio, sell your data, or use it to improve models for other customers. That commitment is reflected in our service agreements with providers and in how the product is built: MCP access is read-only retrieval over your indexed corpus, not a pipeline that exfiltrates your source material.

What can I build with this data?

Rich metadata on every result — speaker, timestamp, programme context, and the exact passage text — lets agents validate quotes before they cite them, assemble multi-source context, and paginate their reasoning across long corpora instead of guessing. Connect once via MCP and every tool in your stack can pull source-verified spoken context on demand. Use-case stories and a testimonial page are coming soon.

How does regest distinguish who is speaking?

Regest uses specialised conversational diarization tuned for dialogue — not generic speaker separation. At upload you can seed likely speaker names; the system then attributes utterances across the recording and lets you reconfirm or edit labels at any time. Every searchable quote carries a speaker label, so agents and humans always know who said what, on record.