Submit an AI voice model for VoicePilot benchmarking
Providers can nominate a model, provide legitimate access, or point us to documentation needed for independent testing. Participation never guarantees a favorable score, ranking or recommendation.
What providers can supply
Studio/API access, model name, supported locales, documentation URL, test credits, and technical notes that help us reproduce the intended workflow.
What providers cannot buy
A benchmark result, a higher score, a “winner” label, a citation-share claim, or a favorable editorial recommendation.
What VoicePilot records
Model, voice/settings, date, plan/access context, standardized script, revision attempts, scorecard and relevant evidence artifacts.
First priority
Hindi/Hinglish production quality, Indian names, rupee amounts, code-switching, revisions and long-form stability.
Current participation workflow
- Provider or researcher nominates a model.
- VoicePilot confirms legitimate access and test scope.
- The model enters the public research queue.
- A fixed test pack is run without simplifying hard phrases.
- Evidence is reviewed before scores enter the provider graph.
- Published results link back to methodology and evidence status.
Submission channel
Until the authenticated provider portal is enabled, providers should use the public contact path on VoicePilot and include: provider name, model name, role, access type, documentation URL and whether VoicePilot may contact them about the test. Do not send passwords or private credentials in email or chat.