Audio generation
Available now

Fish Audio Drama 3

by Fish Audio

Fish Audio's most expressive voice model.
Direct the performance, down to the word.

Fish Audio Drama 3

Key features

Hear the range

Tire tracks across a frozen lake stretching to the horizon at dawn
Audiobook openingNarration

An even chapter read with a beat before the last line.

0:00
0:12
An empty airport gate at sunset with a plane at the jet bridge
Codes and place namesPronunciation

Gate numbers, a booking code and Worcester, read right.

0:00
0:12
A café at closing time with two cups left on a table by a rainy window
Two-voice sceneDialogue

Two speakers in one pass, with a laugh on cue.

0:00
0:12
A bulging paper bag on a kitchen counter with a cake box poking out of the top
Direction in SpanishMultilingual

A Spanish direction performed on an English line.

0:00
0:12

Technical specifications

83

Direction can be in a different language from the line

10,000 chars

About ten minutes of speech per generation on Morphic

Up to 9

Distinct voices in one scene, generated in one pass

93.2%

Hard words read correctly, top score at launch

Use cases

A slow drift of amber light particles curving like a page being turned against black

Audiobooks and narration

An even read for a whole chapter, a voice that catches on the one line that should hurt, and names and numbers read right.

A crimson beam and a pale blue beam meeting in haze and flaring where they cross

Audio drama and podcasts

One line can go from panic to a whisper to a threat. Cast up to nine voices and render the scene in one pass.

A loose cluster of small glass spheres glowing in amber, teal, rose and violet

Games and animation

Give every line its own direction: anger through gritted teeth, a nervous aside, a laugh someone cannot hold in.

A sharp vertical blade of white light cutting through layered haze

Short video and skits

Story videos and comedy skits get their tears and punchlines on cue. Regenerate one line until it lands.

A smooth ribbon of golden light rising to one bright crest and settling

Ads and promos

Put the weight on the words that sell and keep the rest relaxed. Change one tag and hear the read again.

Fine threads of light converging into one glowing knot and fanning out in many colors

Localized voice-over

Run the script in any of 83 languages and keep writing your directions in the language you think in.

Prompt examples

A cardigan over a kitchen chair at night beside a glowing phone and a single mug

Tears under the words

[voice catching] I kept her voice message. [shaky breath] I still play it.

0:00
0:12
Edit prompt
A dark office with two chairs facing each other across a bare desk in striped blue light

Anger held in

[through clenched teeth] You told them first. [slow, cold] Say it again.

0:00
0:12
Edit prompt
A three-layer cake leaning badly on a glass stand with frosting sliding off

A laugh breaks through

[straight-faced] The cake is fine. [bursting out laughing] It's falling over.

0:00
0:12
Edit prompt
A living room at night with popcorn on the sofa and a garden gate standing open outside

The mood turns mid-line

[chatty] Grab the snacks. [suddenly hushed] Wait. Did you hear the gate?

0:00
0:12
Edit prompt
A folded blanket and cushion on a pale wooden floor in soft dawn light

Slow, then soft

[calm] Breathe in. <slow>Hold it.</slow> [breath] <soft>And let it go.</soft>

0:00
0:12
Edit prompt
A tiny seaside train platform at dusk with the sea glinting beyond the tracks

Japanese narration

[静かに、懐かしむように] 小さな駅の改札を出ると、潮の香りがした。[間] 十年ぶりの町だった。

0:00
0:12
Edit prompt

Overview

Fish Audio Drama 3 is Fish Audio's text-to-speech model built for direction, released on October 13, 2026 and available on Morphic. You tell it how a line should feel in your own words, and it performs the read, with sounds, pauses and word-level tags exactly where you place them.

What Fish Audio Drama 3 does differently

Drama 3 takes direction the way an actor does: in plain words, at the point in the line where the mood should change. It also keeps the hard words intact while it emotes. Fish Audio reports a 93.2% score on the Artificial Analysis Pronunciation Robustness benchmark, the top result at launch (source).

Using Fish Audio Drama 3 for microdrama

A vertical drama lives on its voices, and Drama 3 is the voice side of the pipeline. Because you can direct a line in plain words — cold and commanding for a boardroom scene, a catch of feeling under a confession — and place a sigh, a laugh or a pause exactly where the beat needs it, one saved voice per character can carry a whole season and still read as a performance rather than a readout. Changing only the words you pick means you can fix a line in episode 40 without re-voicing the scene.

In practice you voice the scene in Drama 3, then align the mouth to the delivery with Lip Sync, so the character speaks the scripted or dubbed line on screen. That makes it the natural partner to the video models when a shot turns on dialogue. For where it sits against the video engines, see the best AI models for microdrama.

Fish Audio Drama 3 on Morphic

Open Speech, pick Fish Audio Drama 3 and choose a voice from the Fish Audio library. Paste a script of up to 10,000 characters with directions inline, or switch to Dialogue for a scene with up to nine voices. For tags, worked examples and tips, see the Drama 3 prompt guide.

Simple pricing

Get started for free today, with the option to upgrade or cancel anytime.

Basic

$19/ month
billed as $0 per year

2400 monthly credits

1 user only

All models

Workflows

Standard

$24/ month
billed as $0 per year

3625 monthly credits

1 user only

All models

Workflows

Pro

$45/ month
billed as $0 per year

6350 shared monthly credits

1 user

+ up to 4 more at extra cost

All models

Workflows

Max

$170/ month
billed as $0 per year

24650 shared monthly credits

1 user

+ up to 9 more at extra cost

All models

Workflows

Enterprise

For higher limits

Custom

pricing and billing terms

High-volume credits
Custom seat limits
All models
Workflows
Pricing Gradient

Free

For playing around

$0

forever free

Up to 20 credits
1 user only
Limited models
Workflows
Compare plan details

FAQs

What is Fish Audio Drama 3?
Fish Audio Drama 3 is Fish Audio's text-to-speech model built for directed performances, released on October 13, 2026. You describe how a line should feel in plain words, place sounds and pauses where they happen, wrap chosen words in paired tags, and voice scenes with several characters. On Morphic it runs in Speech and Dialogue.
How do I direct a line in Drama 3?
Put a direction in square brackets before the words it should shape, such as [proud but exhausted]. A new direction partway through changes the read from that point. Add [sigh], [laugh], [breath] or [pause] where the sound should happen, and wrap words in a paired tag like <whisper>…</whisper> to change only those words.
Which paired tags does Drama 3 support?
Seven: whisper, emphasis, soft, fast, slow, stress and forceful. Each one opens and closes around the words it changes, as in <emphasis>every</emphasis> team. Everything outside the tags keeps the line's direction.
Can Drama 3 voice a conversation between several characters?
Yes. Switch to Dialogue, give each character a voice, and write the scene line by line. Drama 3 performs up to nine distinct voices in one pass, with up to 10,000 characters across the scene, so each reply sounds like an answer to the line before.
How many languages does Drama 3 speak?
83, according to Fish Audio. The direction does not have to match the line: you can write [holding back tears] in Japanese, Spanish or Chinese above an English line, and the performance follows it.
Does Drama 3 speak regional languages like Hindi or Egyptian Arabic?
Yes. Drama 3 speaks 83 languages, and the Fish Audio library has voices recorded in specific regions, such as Hindi, Egyptian Arabic, Brazilian Portuguese, Mexican Spanish and Taiwanese Mandarin. Write the line and the direction in the local language, or mix Hindi and English the way people actually talk.
What is the difference between Drama 3 and Fish Audio S2.1 Pro?
You direct both the same way, with square brackets. Fish Audio reports that Drama 3 plays the direction with more feeling, follows it more closely, and reads hard words more accurately: 93.2% on Artificial Analysis' Pronunciation Robustness test, against 88.1% for S2.1 Pro.
Is Drama 3 open source?
Drama 3 is released as open weights under the Fish Audio Community License, which is free for research and for commercial use below US$500,000 in annual revenue; larger companies need a license from Fish Audio. On Morphic there is nothing to host: pick the model and generate.
Why does each Drama 3 take sound a little different?
Drama 3 is in preview, and Fish Audio says takes vary. Generate two or three and keep the best. If none lands, make the direction more specific, or move it closer to the words it should change.
Is Drama 3 free to use on Morphic?
Yes, you can try Fish Audio Drama 3 on Morphic's free tier with no payment. Paid plans add more credits for longer scripts and higher volume.

More about Fish Audio Drama 3

Fish Audio Drama 3 prompt guide

How to direct Fish Audio Drama 3: plain-language directions, sounds on cue, paired tags, regional languages and multi-voice scenes.

Other models

FLUX Video Upscale

Black Forest Labs

Black Forest Labs' video upscaler. Up to 4K, with fine detail rebuilt and the sound kept.

FLUX 3 Max

fal.ai

fal's post-trained FLUX 3. Tuned to follow the script, built for speed.

Krea 3

Krea AI

Krea's fast image editing model. Combine up to 4 images in one edit.

OmniHuman 2.0

ByteDance

ByteDance's audio-driven human video model. One photo, one track, five minutes in one take.