The three audio lanes
Each lane has a fixed header column on the left holding its name, its mute button and its generation buttons. Within a lane, overlapping clips are automatically stacked into sub-rows and the lane grows as tall as it needs — no clip ever sits on top of another.
- Diyalog / Müzik / Efekt (dialogue / music / effects) — the names of the three audio lanes. Diyalog carries speech, Müzik the score, Efekt the sound effects.
- Sustur / Sesi aç (mute / unmute, lane) — the speaker icon left of the lane name mutes the entire lane, both in the preview and in the exported film. A muted lane fades out visually but its clips stay editable; press again to bring the sound back.
Tümünü üret: batch-generating pending audio
If a lane has accumulated clips whose audio hasn't been generated yet, you can generate them all in one go. You see the total cost before pressing anything — in the button's tooltip and again in the confirmation dialog; nothing is spent until you confirm.
- Tümünü üret (generate all) — the layers icon in the lane header, left of the + button. Batch-generates every clip in that lane that has no audio yet. Hovering shows the pending clip count and the total price in the tooltip (e.g. "Tümünü üret · 5 · $1.20"). The button is disabled when there are no pending clips or another batch is already running; while generating it turns into a spinner.
- Tüm sesleri üret? (generate all audio? — confirmation dialog) — states "all ungenerated audio clips in this lane will be batch-generated — this spends tokens. Continue?" and shows the total cost in both dollars and tokens. The price comes from the exact same rule that charges your wallet: dialogue is priced by line length, music and effects by the second; crowd (choir) lines and long ambient descriptions on the Efekt lane (which switch to the music rate) are included in the total. Tümünü üret confirms, Vazgeç (cancel) aborts.
- Üretim durumu notu (generation status note) — during a batch run, a line under the lane header says what's happening right now: "Sesler hazırlanıyor (3)", "Ayşe diyaloğu oluşturuluyor", "Müzik oluşturuluyor" and so on. Each clip being generated also shows a small spinner on top of it.
Klip ekle: draft first, then generate
Adding a new clip works draft-first: pressing the button doesn't save anything yet — it opens the clip dialog with a draft placed where the playhead currently sits. If you close the dialog without generating, nothing is created — the draft vanishes. When you press generate for the first time, the clip is saved to the server first and then the audio is generated.
- Klip ekle (add clip) — the + button to the right of the lane header. Prepares a new draft clip at the playhead position on that lane and opens the editing dialog. The button is disabled while the automatic production pipeline is running.
The typical order for a dialogue clip: pick the Konuşan karakter (speaking character) first (or leave it as narrator), write the line, optionally change model and language; the price label below updates live with the text length. Music and effect clips are generated from a description (prompt) and a target duration.
Editing on the strip: move, trim, split
A generated clip shows its waveform on the strip, and most editing happens right there. Clips never overlap and never leave their own lane.
- Klip (clip, on the strip) — a generated clip shows its waveform, character badge, duration and label; an ungenerated clip is faded and italic (tooltip: "Henüz üretilmedi" — not generated yet). One click selects it (orange outline and edge handles appear), double-click opens the editing dialog. A muted clip fades heavily but can still be moved and edited.
- Klip sürükleme (dragging) — horizontal dragging moves the clip in time; vertical dragging moves it between sub-rows within the same lane (including a new empty row at the bottom). You can't drop it onto another clip — it settles into the nearest free spot, snapping against its neighbour. While dragging, a live seconds value reads out above the cursor; the clip snaps to a grid that scales with zoom, and holding Alt disables the grid for millisecond-precision placement.
- Sol kenar tutamacı (left edge handle) — trims or re-opens the clip's start; the right edge stays put while the in-file start of the audio shifts. It can't go past the audio's true start, the previous clip, or the beginning of the timeline.
- Sağ kenar tutamacı (right edge handle) — changes the clip's duration. It can't extend past the audio's real length (no silence is added) and can't push into the next clip in the same row.
The right-click menu offers five actions:
- Düzenle (edit) — opens the clip's editing dialog (same as double-click). The menu closes with Escape or a click elsewhere.
- Çoğalt (duplicate) — creates a copy of the clip.
- Oynatma çizgisinden böl (split at playhead) — cuts the clip in two where the vertical playhead sits — like the razor in video editors. It only appears on clips with generated, non-choir audio; if the playhead isn't inside the clip and far enough from both ends, the item is disabled with the hint "Önce oynatma çizgisini bu klibin içine getirin." (bring the playhead inside this clip first).
- Sustur / Sesi aç (mute / unmute, clip) — mutes only this clip: it stays in place with its takes and settings intact, but doesn't play in the preview and is skipped on export. Unmute from the same menu.
- Sil (delete) — deletes the clip. A confirmation appears first: "this item will be permanently deleted. Continue?" — Sil confirms, Vazgeç cancels.
The clip dialog: listening and mixing
Double-click (or right-click → Düzenle) opens the clip dialog. Here you listen to the clip and set its volume, fades and stereo position. There's no separate save step — every change is saved the moment you make it.
- Konuşan karakter (speaking character) — at the top of the dialog, Diyalog clips only. Picks whose voice speaks the line; the "Karakter seç" (pick a character) dropdown lists every character plus "Anlatıcı (karaktersiz)" (narrator, no character). A new clip opens with no speaker.
- Oynatıcı (player) — on generated clips: a round play/pause button, a click-and-drag waveform bar and a duration readout. It plays only the audio range the clip actually uses (a trimmed or split piece shows its own slice) and applies the volume and fades live. If no audio has been generated, a "Henüz üretilmedi" (not generated yet) box appears instead.
- Sessizlik uyarısı (silence warning) — if generation returned an empty or near-silent recording, a warning under the player says "this audio is nearly silent — try regenerating." Since waveforms are normalized, a silent clip would otherwise look full.
- Ses seviyesi sürgüsü (volume slider) — a vertical bar; top is loud, bottom is silent, with the percentage shown below. The speaker icon above it mutes/restores with one click. The setting saves instantly; preview and export use the same level. Default mix: dialogue 100%, music 40%, effects 65%.
- Açılma / Kısılma (fade-in / fade-out envelope) — drag the shaded box's top-left handle inward to define a soft fade-in at the start, the top-right handle for a fade-out at the end; their durations read out in seconds under the box. The two handles can't cross; the shaded shape is the actual envelope you'll hear, and export applies the same ramps.
- Sahne konumu (pan) — an L—R horizontal slider setting where the clip sits in the stereo field (L100…C…R100), with a light magnet at center. Dialogue clips are normally auto-positioned from the character's placement in the shot; touching the slider pins the value.
- Oto (auto — pan reset) — the small badge right of the pan slider, Diyalog only. Releases a manually pinned pan value and re-derives the position from the shot's blocking.
- Tamam (done) — in the dialog's fixed bottom bar; closes the dialog. Everything is already saved. An ungenerated draft closed without ever pressing generate is discarded without being created.
Takes: no attempt is ever lost
Regenerating never deletes the old audio; each attempt is added to the list as a new take. Switching back to any take is free and instant.
- Sürümler (takes/versions) — the list under the player (when there's more than one attempt). Clicking a row makes that take active: the clip plays that audio and its duration adapts to it. Each row shows the generation text, the duration and the token cost of that attempt.
- İndir (download, take row) — the download icon that appears on hover downloads that take's audio as an mp3; a spinner shows in the row while downloading. Free.
- Sürümü sil (delete take) — the trash icon that appears on hover permanently deletes that take. An in-dialog confirmation asks "this audio take will be permanently deleted. Continue?" (Vazgeç / Sil). The last remaining take can't be deleted; if the active take is deleted, one of the others becomes active. The action is written to the undo history.
Üret / Yeniden üret: the Metinden tab
The collapsible generation panel at the bottom of the dialog defines how the audio gets made. The text, model and duration here affect only the next generation — editing them doesn't change the existing audio.
- Üret / Yeniden üret (generate / regenerate — collapsible section) — the panel header that opens and closes the generation settings. On a clip with no audio it opens expanded as "Üret"; on a clip with audio it starts collapsed as "Yeniden üret".
- Metinden / Canlandır (from text / perform — tabs) — on Diyalog clips with a character only: two ways to produce the same line. Metinden has the AI voice read the text you wrote; Canlandır lets you record the line in your own voice and convert it into the character's timbre. The dialog always reopens on the Metinden tab.
- Konuşulacak metin… (line to speak) — the actual line to be voiced on Diyalog clips. Exactly what you type is spoken; you can add square-bracket tags for delivery:
[shouting],[whispers],[angry],[sad]… The tags are highlighted in the field and passed through verbatim. Editing the text clears older performance marks; the next generation speaks exactly what's written. - Sesi tarif et… (describe the sound) — on Müzik and Efekt clips, a plain-words description of the music or effect to generate. The price label and the generation use this text.
- Model — picks this clip's audio engine. Diyalog: Oto, runime Ses, ElevenLabs; Müzik: Oto, runime Müzik, ElevenLabs Music; Efekt: Oto, runime SFX, ElevenLabs SFX. Options list a version badge and a consistency star rating; models switched off by the admin don't appear. If you pick nothing, the film's configured engine for that lane is used.
- Dil (language) — Diyalog only; sets this line's voice-over language independently of the film's language — a character can speak another language. Default is "Otomatik (film dili)" (automatic, film language); the list offers Türkçe, English, Español, Deutsch, Français, Italiano, Português, Русский, العربية, 日本語, 한국어, 中文, हिन्दी and Suomi.
- Süre (duration) — Müzik/Efekt only, a clock-icon number field: the target length of the next generation, in seconds. Music is clamped to 3–600 s, effects to 0.5–22 s. Changing the value doesn't touch the existing audio; after generating, an existing trim is kept, and if the result is shorter, the clip shortens. Dialogue duration is never set by hand — the real length of the generated speech is adopted.
- Fiyat etiketi (price label) — right of the model/duration row; before you click anything it shows what the generation will cost, in both dollars and tokens (e.g. "$0.34 · 1.200 token"); if the cost is zero it says "Ücretsiz" (free). Dialogue is priced by text length, music and effects by requested seconds; the number updates live as text or duration changes and matches the wallet charge exactly.
- Üret / Yeniden üret düğmesi (generate / regenerate button) — the icon button embedded in the bottom-right of the text field generates the audio: a star icon for the first generation, a loop icon for regeneration; while running it turns into a spinner and disables. On a draft clip it saves the clip first, then generates. Generation spends tokens; the amount is on the price label beside it.
Canlandır: perform the line in your own voice
The Canlandır (perform) tab does record-to-voice conversion for a single line: delivery, emotion and pacing come from you, the timbre from the character. (This is not a voice clone — assigning a character a permanent voice clone lives on the character-editing side.)
- Kendi sesinle canlandır (perform with your own voice — record) — the round red microphone button starts the recording flow: first a 3·2·1 countdown with "Hazır ol…" (get ready — the countdown isn't recorded), then recording begins, with live level bars and a time counter. If microphone permission is denied, a warning says "Mikrofona erişilemedi. Tarayıcı izni gerekli." (couldn't access the microphone — browser permission required).
- Duraklat / Devam et / Durdur (pause / resume / stop) — while recording: Duraklat holds the recording (the mic stays open), Devam et resumes, the red Durdur finishes and shows a preview player of the raw recording.
- Yeniden kaydet (re-record) — discards the current recording and returns the tab to the start. Disabled while conversion or applying is in progress.
- Karakter sesine dönüştür (convert to character voice) — converts your recording into the character's timbre. Shows "Dönüştürülüyor…" while running; on failure, "Dönüştürme başarısız. Tekrar dene." (conversion failed, try again). This step is paid and billed by recording length — no separate price label is shown on this tab. On a draft clip, the clip is saved first.
- Uygula (apply) — under the converted preview headed "character voice (preview) — if you like it, apply". Adds the converted audio to the clip as a new take and makes it active; it appears in the Sürümler list with a "🎙️ kayıttan" (from recording) label. If you don't like it, start over with Yeniden kaydet — the clip doesn't change until you press Uygula.
Costs and things to watch
Three actions spend tokens, and all but one show the amount before you press anything:
- Single-clip generation/regeneration — the price sits on the Fiyat etiketi (price label) next to the generate button, in dollars and tokens, before you click. Dialogue is priced by text length, music and effects by requested seconds. Hidden multipliers are already in the number: crowd/choir lines cost a multiplied rate, and ambient descriptions of 10 seconds or more on the Efekt lane are billed at the music rate.
- Per-lane Tümünü üret (generate all) — the total price appears both in the button's tooltip and in the confirmation dialog; nothing starts without your confirmation.
- Karakter sesine dönüştür (convert to character voice) on the Canlandır tab — measured and billed on the server by recording length; this tab shows no price label, so know that it's a paid step.
Free: picking a take or returning to an old one, downloading takes, moving, trimming, splitting, duplicating, muting, and volume–pan–fade adjustments.
Watch out: a draft clip closed without generating is never created. The last remaining take can't be deleted; deleting a take is permanent and asks for confirmation, as does deleting a clip. Lane and clip mutes affect both the preview and the export. Near-silent generations trigger a warning in the dialog — try regenerating. Clips never overlap and always stay on their own lane.