Skip to content

PODCAST ARCHIVE LICENSING

License your podcast archive for AI training

Keep your show and ownership. Explore whether a documented subset of your original recordings fits a defined AI use, with the source, permissions and commercial terms reviewed before delivery.

Updated November 5, 2026 · By Fiund

Start with the originals and the intended task.

An interview archive can contain natural conversation, interruptions, laughter, speaker changes and different acoustic conditions. Those features can be useful when the proposed model task needs them. The value of a collection depends on the requirements it can actually meet, rather than the fact that a show is public or popular.

A speech-recognition project may need accurate transcripts and a defined treatment of punctuation, overlap and non-speech sounds. Speaker diarization concerns who spoke when. Audiovisual tasks need a documented time relationship between sound and picture. Source audio, a model-generated label and a human-reviewed annotation are different deliverables. Record which are already available and which would require preparation.

Match the source to a specific AI task
TaskUseful source informationA limit to preserve
Speech recognitionOriginal audio, language, acoustic conditions and transcript conventions.A rough transcript is not automatically a checked reference.
Speaker separationSpeaker turns, overlap and original isolated channels when available.A mixed track does not contain separately recorded source channels.
Audiovisual understandingCamera originals, audio linkage, timing and on-screen context.Published edits can omit or change the original time relationship.
EvaluationReviewed examples, task definitions, held-out episodes and known limitations.Closely related copies should not leak across training and evaluation splits.

Build one episode record before estimating the archive.

Use an internal episode ID to connect the mix, raw audio, separate speakers, camera files and transcript. Preserve the original filename, version and location in your own inventory. Mark source and processed versions so an evaluator can see what changed.

The RSS catalog audit lists the episodes and metadata present in a supplied feed. It flags missing durations and repeated identifiers. It cannot find recordings absent from the feed or verify the master files. Use the Media Dataset Inspector to read local file properties, then reconcile those results with your episode and permission records.

Count unique recorded conversation separately from simultaneous audio tracks and camera streams. A 40-minute session with two microphones and two cameras is 40 minutes of activity with several source streams. It is not automatically 160 unique conversation minutes. Keep any stream-hour quantity labeled so a quote or delivery can be compared fairly.

Map the rights to the material actually proposed.

Identify who controls the recording and any producer, employer, client, network or co-host interest. Review the intended AI use against the actual source and participant permissions. A release written for ordinary podcast distribution should not be assumed to answer every later use.

Mark music, advertisements, stock footage, screen shares and other embedded work. Identify relevant time ranges and the reason for exclusion or review. Sensitive discussion, private details and off-record material need separate attention across audio, video and transcripts. Access to a public feed or a clean text transcript does not settle these questions.

Keep unknown status visible. An episode can have a located master and an unresolved guest grant at the same time. The archive readiness guide works through a fictional episode with music and ad ranges and explains why the remaining runtime still needs permission review. Use the private readiness checker to turn those gaps into follow-up tasks.

Understand the standard agreement before participating.

The published standard contributor agreement is the governing source for this summary. Viewing a guide or completing a preparation checklist does not accept that agreement. Read the complete terms and any applicable written addendum before deciding whether to participate.

  • Ownership stays with you. Fiund licenses permitted uses of submitted content rather than acquiring your show or feed.
  • Buyer licences are non-exclusive. Fiund has no authority under the standard agreement to grant exclusive, first-look or first-refusal rights.
  • Licensing uses standing authority. Fiund negotiates and closes buyer licences within the agreement’s scope without asking you to approve or countersign each transaction.
  • Standard compensation is defined. For new submissions, the standard share is 50% of attributable net receipts after permitted documented expenses. Prior agreed shares for existing content are preserved. Different terms require an accepted written addendum. A headline buyer payment is not the same as net proceeds attributable to one contributor.
  • Withdrawal concerns future licensing. Existing buyer licences can continue for their stated terms under the agreement. Review the withdrawal procedure and notice requirements.
  • Identity replication is prohibited. Standard licences cannot authorize a synthetic replica, clone or voice or likeness model intended to imitate an identifiable person, or use identity to imply endorsement.

Read how payments work for the payment process. Rates, accepted material and any preparation work depend on a real scope. An assessment, an expression of buyer interest and a completed payment are distinct events.

Use a small, documented first sample.

  1. Describe the archive. Supply the show link, approximate volume, languages, recording formats, transcript status and what you know about permissions. Keep estimates labeled.
  2. Review fit and open questions. Fiund examines the description and can ask for further source or rights details. A missing field should remain unknown until supported.
  3. Follow the agreed sample process. If a review is appropriate, identify the exact files, recipient and evaluation purpose before transfer. Keep originals and evidence linked.
  4. Read and accept the applicable terms. The contributor agreement and any addendum define the relationship, allowed use, authority and compensation.
  5. Prepare an accepted scope. A buyer match, licence, delivery and payment proceed only when the actual requirements and terms are satisfied. Joining does not guarantee demand or a sale.

Fiund can assess existing recordings and identify preparation that needs separate scoping. Transcription, alignment, labeling, redaction and human review should not be described as complete unless they have been performed and checked for the selected files.

Keep archive licensing separate from audience revenue.

You can evaluate a licensing opportunity while continuing to publish and pursue sponsorships or memberships, subject to your existing commitments. The monetization guide helps compare those audience-based models. The revenue calculator models ads and memberships and assumes no AI licensing revenue.

For a comparison of public licensing programs, read the Troveo review, Scale AI review and Appen versus Defined.ai comparison. A provider’s published program does not establish acceptance for your particular archive.

Frequently asked questions

How much can I earn from licensing a podcast?

There is no universal rate or guaranteed sale. Fit, usable volume, source quality, permissions, buyer requirements and preparation affect the opportunity. Standard compensation is defined in the contributor agreement, including the definition of Net License Revenue and payment conditions.

Can I keep publishing and monetizing my show?

You retain ownership. Standard buyer licences are non-exclusive. Check existing network, sponsor, platform and licensing commitments, and read the actual contributor agreement before participating.

Do I approve each buyer deal?

The standard agreement gives Fiund standing authority to negotiate and close buyer licences within its scope. It does not require a separate contributor approval or signature for each deal. You can withdraw content from future licensing as described in the agreement.

Does licensing authorize voice cloning?

No. Standard terms prohibit a synthetic replica, clone or voice or likeness model intended to imitate an identifiable person, and prohibit using identity to imply endorsement. Buyer licences must carry that restriction.

Do I need to upload my entire archive?

No. Start with a description and show link. Identify the originals and permission records you have, and mark gaps. Any later sample review or transfer should follow the agreed scope and process.