Your voice stays yours

TalkCopyPaste never owns your audio or transcription keys. You choose the AI. You control the data.

Download TalkCopyPaste See how it works

The problem with how voice apps handle your data

Most voice apps process your audio on their servers. You speak, they receive, they transcribe, they (maybe) delete.
Privacy policies are long, change frequently, and rarely explain exactly what happens to your recordings.
Users don't know who has access to their voice data—the app company, their cloud providers, their AI vendors.
Trust is assumed, not earned. You're asked to believe a privacy policy instead of verifying the architecture.

The Core Principle

You should control the AI that hears your voice.

You choose the provider. Groq, OpenAI, or a local Whisper model—your call, your relationship, your terms.

You use your own API key. Your account, your usage, your billing. No middleman.

Nothing routes through TalkCopyPaste servers. Audio goes directly from your computer to your chosen provider.

We literally cannot access your recordings. Not "we promise not to"—we architecturally cannot.

Ready for privacy you can verify?

Setup takes 60 seconds. Bring your own key, keep your own data.

Download Free Get a free Groq API key

How BYOK changes your relationship with voice AI

Privacy through architecture, not promises

One key, full control

Paste your API key once. From then on, your audio flows directly to your provider. Change providers anytime by swapping the key.

No data middleman

Traditional apps receive your audio, process it, then send results. With BYOK, we're not in the data path at all.

We can't see your data

This isn't a policy decision—it's architecture. Your audio never touches our servers, so we have no access to it.

Use local models

Run Whisper locally and your voice never leaves your machine. Perfect for sensitive work or air-gapped environments.

Switch providers anytime

Not happy with Groq? Switch to OpenAI. Want to try a new model? Just update your API endpoint. No vendor lock-in.

Compliance-friendly

For regulated industries, BYOK means you control the data flow. Easier audits, clearer accountability, simpler compliance.

Why BYOK scales better over time

Control compounds. Lock-in doesn't.

Not tied to a single vendor. When better models emerge, you can switch without changing apps or workflows.

Users aren't locked into our pricing decisions. If we raised prices (we won't), you'd still just pay your API provider directly.

Trust scales as usage increases. The more you use TalkCopyPaste, the more confident you become—because the architecture doesn't change.

Privacy improvements happen automatically. When providers improve their privacy practices, you benefit immediately.

Common questions about BYOK privacy

Is BYOK harder to set up?

No. It takes about 60 seconds. Visit Groq (free) or OpenAI, create an API key, paste it into TalkCopyPaste settings. That's it. One-time setup, then you never think about it again.

What if I don't want to manage API keys?

We understand. But consider the alternative: your voice recordings stored on a company's servers, processed by unknown systems, retained for unknown periods. Managing one API key is a small price for knowing exactly where your data goes.

Does Groq or OpenAI store my audio?

Check their privacy policies for current details. Groq states they don't use API data for training. OpenAI offers a zero-retention API option. Either way, you're dealing directly with your chosen provider—not a middleman who adds another layer of data handling.

Can I use a completely local model?

Yes. TalkCopyPaste supports any OpenAI-compatible API, including local Whisper implementations. Run Whisper on your own machine and your audio never leaves your computer. Zero cloud, zero API costs, maximum privacy.

Why don't you just offer your own API?

Because then we'd have to handle your audio. We'd need servers, storage, security audits, compliance certifications. More importantly, you'd have to trust us. With BYOK, you don't need to trust us—your audio never touches our infrastructure.

Is this more expensive than apps with built-in transcription?

Usually cheaper. Groq offers a generous free tier (enough for 2+ hours of dictation daily). Even paid usage is typically $0.04/hour. Subscription apps like Wispr Flow cost $144/year regardless of usage. You only pay for what you use.