Voice mode
Talk through your next move with one AI mentor or your whole team. Same founder research and chat context, spoken in a cloned voice.
Ask a follow-up, interrupt, or switch from English to French. Voice mode is optional; text chat works without Fish Audio.
Choose Call in the chat header. AI simulation, not a call with the real founder.
Tap a founder to hear the short public excerpt their AI voice is cloned from. Prepared recordings are selected automatically; other mentors use a labeled generic AI voice. Generated replies are AI simulations, not statements by the real founders.
Everything happens inside the chat. No Sauce account needed: you bring your own AI provider and Fish Audio keys, and setup walks through each one.
Choose Call in a mentor’s chat. The first step connects the LLM key that writes the replies from the mentor’s research and your chat context. Existing connections are reused.
A separate Fish key turns those replies into speech. Choose S2.1 Pro Free for free speech generation under Fish’s current terms. Fast transcription also needs available Fish API credit.
Get your Fish Audio API key ↗The host and invited mentors each get a prepared founder voice where available, or a labeled generic AI voice. Change any voice, upload a clear 10–30 second excerpt, or connect a Fish voice ID.
Choose Fast for live transcription or Local for free on-device transcription. Review the estimated costs, press Start call and allow microphone access. Your microphone stays off until you start.
Use a current browser with microphone support. Headphones keep the mentor’s audio from being picked up as your speech. You can also type during a call. Invited chat mentors join the call, up to eight people, each with a separate voice.
Choose Invite mentors inside the call to add people at any time, or invite them in your chat first. The host answers first. Follow-ups stay with the current mentor; say another name or select their tile to switch.
Product context in the call toolbar lets you edit your details, switch products or clear the selection. Changes apply to your next question. Opening it pauses your microphone and stops the current reply; returning restores your previous mute setting.
How should I launch?
One mentor answers in their own voice.
Julia, what about creators?
Julia answers and takes the follow-ups.
Everyone, review my plan.
Each mentor takes a short turn. Interrupting cancels the remaining replies.
One Fish key and one transcription session for the group. Each mentor who answers adds AI usage and, if you selected paid speech, speech usage. Setup shows estimates for one reply and the whole team.
You create your own Fish account and bring your own API key. You pay Fish Audio directly. Sauce receives none of these payments and adds no markup. Your AI provider bills its replies separately.
Local transcription with S2.1 Pro Free
$0in Fish costs
No Fish top-up needed. The first call downloads a transcription model, and replies wait for a slightly longer pause after you finish speaking.
Live transcription with S2.1 Pro Free
$0.36per hour connected
About $0.06 for ten minutes connected, billed per second, silence and mute included. Requires available Fish API credit: the free speech model does not pay for transcription. Microphone audio goes through Sauce to Fish. Add API credit on Fish ↗
Fish advertises free S2.1 Pro Free access through November 30, 2026, subject to fair use and its terms, with no latency guarantee. The offer may change. Paid S2.1 Pro is optional, currently $15 per million UTF-8 bytes of speech text, and Sauce never switches your call to it automatically. Fish’s website subscriptions and saved-voice allowances are separate from API pricing.
Setup shows an estimated AI reply cost, and the call shows a running USD estimate for replies, speech and transcription, using reported token counts when available. Interrupted calls, unknown prices and account discounts can differ from the provider invoice. GimmeTheSauce is not affiliated with or endorsed by Fish Audio. We recommend it because its API currently offers free speech generation.
Checked October 31, 2026. Official API prices ↗/Current free-tier terms ↗
Hold the mentor’s voice and resume from the same word.
Speak again to interrupt, or choose “Let me jump in.”
French input gets a French reply. Switch languages or ask for another one. Recognition and pronunciation can vary.
Replies appear as they’re spoken and stay in your current chat. Toggle the transcript on or off while talking.
Turns your microphone off. Fast transcription still bills connected time while muted.
Hang up, close the call or close the tab to end it. Connections end after about 12 minutes; start another call to continue.
Providers apply their own processing and retention policies; Fish says free-tier requests may be used to improve model quality. Already-submitted requests may still be billed after you hang up.
Read the full data and key guideFast transcription and the paid S2.1 Pro speech model need available Fish API credit. If your API balance is empty, add credit on Fish’s API Billing page ↗ before calling. API usage is pay as you go; no Fish subscription is required, and a website subscription is separate from API billing. Local transcription plus S2.1 Pro Free works without a top-up under the current free-tier terms. Your AI provider’s usage is billed separately either way.
No. You’re talking to an AI simulation informed by public research. Cloned speech is generated, and does not mean the founder said those words, is participating, or endorses Sauce.
No. Prepared and uploaded samples use instant reference cloning: the clip is sent with each speech request, without creating a saved model in your Fish library. Each mentor can have its own reference. Fish’s API access, fair-use and request limits still apply.
10–30 seconds of clear speech from one speaker, without music or overlapping voices. You can trim a short audio or video recording in setup, preview it and provide the exact words. If you leave the transcript blank, Sauce transcribes the clip locally.
Voice calls and Fish-key setup work on the Sauce website. MCP and the portable skill provide mentor research to your assistant; they don’t access this browser’s microphone, voice settings or keys.
Open any mentor’s chat and choose Call in the header. AI simulation, not a call with the real founder.