Audio Tools · Dialogue Scripts · 12 Languages
Paste a dialogue script, give every speaker their own voice
Paste a script and it splits into lines, then gives each speaker name its own voice. Every line has a role dropdown, so you can move any line to a different speaker.
Paste the whole script at once. Each block of dialogue becomes its own segment — split on blank lines, or on line breaks when there are no blank lines — and nothing gets dropped. You can also start from an empty segment and write it here one block at a time.
Names written as "Alice:" or [Alice] at the start of a paragraph are picked up and turned into roles. If the script has no name prefixes, every segment lands on role 1 and you assign the speakers yourself.
Open the voice picker for each role and choose a voice, plus an emotion if you want one. A name that matches a voice in the library is bound to that voice already; the other names get voices in the order they first appear.
Use the role dropdown on any segment to move a line to a different speaker, and insert a pause or a laugh where the read needs one. Then generate and export the audio with SRT subtitles.
It reads a dialogue script aloud with a different voice for each speaker, so one script becomes a conversation. You paste the script, assign a voice per role, and export a single audio file.
It looks for a speaker name at the start of each paragraph — "Alice:" with a half-width or full-width colon, or [Alice] in brackets — and names can be up to 12 characters. The speaker split is applied only when at least 60% of paragraphs carry a name and at least two different names appear; otherwise everything goes to one role and you assign the speakers yourself.
Up to 10 roles, 200 segments, and 20,000 characters in total. You can generate in 12 output languages.
Yes. Every segment has a role dropdown, so you can move any line to a different speaker at any point. Right after a paste an undo also appears once — it can revert the paste, or just undo the speaker split and keep the text.
Yes, in two ways. Each voice can carry an emotion, set inside the voice picker; the default is automatic, where the read is performed from the text itself. You can also insert tone tags at the cursor: laugh, chuckle, sigh, inhale, throat-clear, hesitation, and pauses of 0.5s, 1s, or 2s. Tag support depends on the voice — pauses work with more voices than the onomatopoeia tags, which are not available on Japanese voices, and a tag the selected voice cannot handle is flagged in the editor and blocks submission until you remove it.
Yes. Export the finished audio file, and export SRT subtitles alongside it when you need captions.
You wrote the episode as a back-and-forth between a host and a guest. Give each one a voice and both parts come out sounding like two people.
Voice a practice conversation so the learner hears two distinct speakers. Export the SRT alongside the audio for on-screen text.
Give each character in a scene its own voice and emotion, then drop in a sigh or a one-second pause where the writing needs a beat.
Short two-character scripts for a promo or a social clip. Write it in the editor, hear it read back, and adjust the roles until the exchange lands.