Voice-Memo Studio

In the Voice-Memo Studio you create spoken messages in an AI voice. You type or dictate the sentence, choose a profile and voice, and generate. Phrases that keep coming back can be saved as a memo — then the voice is there instantly, without generating it again. Every play button is a toggle (press it again to stop), and closing the window stops playback. A generation in progress keeps running and is ready the next time you open the studio. The studio is its own window — the main window stays open next to it.

You don't choose a technology, you choose a tier: · chatFlow Instant (25 credits) — your voice, generated directly. The fastest way. · chatFlow Design (25 credits) — a voice described in words, then given the timbre of your trained model. · chatFlow Signature (35 credits) — your voice, then run through your trained model: the best version. Design and Signature require a trained voice model. You set the tier as the default in the profile under “GENERATION”. For a single sentence you can choose a different one in the “Tier:” row; the price is shown on every tier.

With chatFlow Design, the “Style:” row above the AI sentence holds your own style list — twelve styles to start with (warm, casual, calm, cheerful, seductive, confident, whispering, professional, narrating, excited, tired, apologetic). One click selects a style, a second click deselects it; the arrow on the right shows all of them, and “+” creates a new one. Right-clicking a style opens “Edit …” (name and instruction) or “Delete” — deleting happens without a confirmation, and for 12 seconds you can bring it back. Style instruction AND description must be written in ENGLISH, otherwise they don't take effect; the spoken language isn't affected by this. Reliably controllable are gender, age, pitch, tempo, emotion, timbre and use case — accents and celebrity impressions don't work. Who is speaking is described below the AI sentence in the “Description (who is speaking)” field; how they speak is set by the style. With Instant and Signature, instead of the style there's “Delivery” (brisk or calm); with Design it has no effect and is therefore hidden. For all tiers: “Variation” determines how much two runs differ from each other, and the “Seed” makes a result repeatable.

Below the tracks are “Similarity” (a slider) and “Sibilants” (three levels: natural, clear, very clear): “Similarity” determines how close the result is to your target voice, “Sibilants” protects s and sh sounds from discoloration. Both only apply to the tiers with a trained model. You set the default in the profile. In the studio, the value applies to the single sentence: it follows the default until you change it there.

The “Match the transition” card bundles the filters that make the transition between the AI voice and your own recording unnoticeable (timbre, volume, tempo). Each filter has its own row; there you turn it on or off or choose a level, and some also have a slider. If a filter costs credits, the price is shown on its row. This card was briefly removed on 4 August 2026 — it's back, because the filters kept working even without a switch.

Above the audio track, the individual words appear as tiles. Click one and you can set its tempo separately — that way you can stretch or tighten a single word without touching the whole sentence. The word timings come from the transcription of the recording.

In the sound library you cut short sounds from your own recordings — laughs, sighs, breaths — and later insert them into generated sentences. Sounds are always tied to ONE voice: someone else's laugh in your own voice stands out immediately. That's why they're only available with chatFlow Instant and Signature: with Design, a described voice that isn't yours, no sounds are inserted, and the style row takes their place. You open the library from inside the Voice-Memo Studio, per profile. If you remove a profile in the Voice-Memo Studio, its sounds go with it; the confirmation tells you beforehand how many. To delete sounds on their own, use the “Sounds” tab.

In both studios: Space plays and pauses, comma rewinds (press again = faster), period stops. In the Voice-Clone Studio, Backspace in Flow mode deletes the ticked clips or the clip that has focus; in the Voice-Memo Studio, Backspace does nothing. Clicking the tempo display cycles the playback speed: 1.0 → 1.25 → 1.5 → 2 → 3 → 0.5. There's also reverse playback with sound.