Scenema Team

Scenema Audio: Zero-Shot Expressive Voice Cloning and Speech Generation

What is Scenema Audio?

Scenema Audio is an audio diffusion model extracted from LTX 2.3. It generates speech with emotional acting, pacing, breath control, and sound effects from a text prompt. It is not speech synthesis. It is performative audio generation.

Zero-Shot Expressive Voice Cloning

Every voice cloning system available today locks emotional range to whatever the reference recording contains. If the reference is calm, the output is calm.

Scenema Audio separates voice identity from emotional performance. The model generates the performance from the prompt, then transfers the voice identity from a short reference clip. The reference only needs to be a few seconds of clean speech. The emotion comes from the generation, not the reference.

Reference audio: Obama (16 seconds, emotional speech)
Reference audio: Tarkin (16 seconds, calm and commanding)
Reference audio: Australian woman (8 seconds, comedy set)

Obama: Expressive Voice Clone

Obama's voice identity transferred from the reference clip above onto an emotionally charged road rage performance. The reference contains no anger. The anger comes entirely from the generation stage.

Show prompt
Voice Gravelly male voice, fast talking, rough. Scene Absolute silence.
He lays on the horn, spit flying, veins in his neck popping as he completely loses it
I got places to be you stupid son of a bitch! What the fuck are you waiting for?! ...a written invitation?!

Tarkin: Expressive Voice Clone

A clipped British voice identity transferred onto a man losing his composure with joy. The reference is calm and commanding. The excitement comes entirely from the generation stage.

Show prompt
Voice Male, late 50s. Clipped aristocratic British accent. Losing his composure with joy. Scene Absolute silence.
He stares at the lottery ticket, hands shaking, voice rising from stunned disbelief to uncontrollable excitement
Those are my numbers. Those are my actual numbers! Oh my god! Margaret! Get in here right now!

Voice Clone: Accent Transfer

The reference is Australian, but the voice description drives the accent of the output. Here the prompt says 'Australian accent' and you get a blend. Change the prompt to 'American accent' and you get the same person speaking American English. This is useful when you want a voice identity performing in a different accent than their reference.

Show prompt
Voice Female, late 20s. American accent. Bubbly, laughing, barely able to get the words out. Scene Absolute silence.
She is already laughing before she even starts talking, barely holding it together
Hahaha! Oh okay okay so... a man walks into a library and asks for books about paranoia. The librarian whispers... they are right behind you. Hahahaha! Gets me every time. Haaaha!

Any voice can perform any emotion, even if that voice has never been recorded in that emotional state.

Polyglot Voice Cloning

From a single English reference clip, the model generates native-sounding speech in any supported language while maintaining the speaker’s identity.

Obama in Spanish

Obama's voice identity speaking native Castilian Spanish. The reference clip is English.

Show prompt
Voice Male, early 60s. Deep baritone. Measured cadence, warm authority. Native Spanish fluency, Castilian accent. Scene A podium, formal address. Absolute silence. Language ES
He pauses, then addresses the crowd in Spanish
Hoy nos reunimos no como naciones separadas, sino como una sola comunidad. Los desafios que enfrentamos no conocen fronteras. El cambio climatico, la pobreza, la injusticia. Estos no son problemas de un solo pais. Son problemas de todos nosotros.

Obama in French

Obama's voice identity speaking Parisian French. Same English reference clip.

Show prompt
Voice Male, early 60s. Deep baritone. Measured cadence, warm authority. Native French fluency, Parisian accent. Scene A quiet conference room, diplomatic setting. Absolute silence. Language FR
He leans forward, speaking directly
La liberte n est pas un cadeau que l on recoit. C est une responsabilite que l on porte chaque jour. Et cette responsabilite, mes amis, elle commence ici. Dans cette salle. Avec chacun d entre nous.

Obama in German

Obama's voice identity speaking Hochdeutsch. The accent adapts to each language natively.

Show prompt
Voice Male, early 60s. Deep baritone. Measured cadence, warm authority. Native German fluency, Hochdeutsch. Scene Berlin, outdoor address. Absolute silence. Language DE
He looks out over the crowd
Die Mauern, die uns trennen, sind nicht aus Stein. Sie sind aus Angst, aus Misstrauen, aus dem Glauben, dass wir verschieden sind. Aber ich sage Ihnen heute: Wir sind nicht verschieden. Wir waren es nie.

Obama in Italian

Obama's voice identity speaking Italian. 100% transcription accuracy.

Show prompt
Voice Male, early 60s. Deep baritone. Measured cadence, warm authority. Native Italian fluency. Scene A formal gathering. Absolute silence. Language IT
He addresses the audience
Oggi ci riuniamo non come nazioni separate, ma come una sola comunita. Le sfide che affrontiamo non conoscono confini. Il cambiamento climatico, la poverta, l ingiustizia. Questi non sono problemi di un solo paese. Sono problemi di tutti noi.

Obama in Swahili

Obama's voice identity speaking Swahili. Less training data for this language results in some phonetic drift, but the voice identity and cadence transfer clearly.

Show prompt
Voice Male, early 60s. Deep baritone. Measured cadence, warm authority. Native Swahili fluency. Scene A formal gathering. Absolute silence. Language SW
He addresses the audience
Leo tunakusanyika si kama mataifa tofauti bali kama jamii moja. Changamoto tunazozikabili hazijui mipaka. Mabadiliko ya hali ya hewa, umaskini, dhuluma. Hizi si matatizo ya nchi moja. Ni matatizo yetu sote.

Voice Design

No reference audio needed. Describe the voice in the prompt and the model generates it from scratch.

Old Male Storyteller

English. Gravelly warmth. The kind of voice you hear around a dying fire.

Show prompt
Voice Male, mid 60s. Deep baritone with gravel. Slight Southern American inflection. Worn but warm. The voice of someone who has seen too much and chosen kindness anyway. Nostalgic, firelight cadence. Scene Fireside, night, crickets in the distance
He settles into his chair and stares at the fire
There was a summer, back when the river still ran clear, when my father took me out past the property line and pointed at the stars. He said, boy, every one of those is a story somebody forgot to write down.
He smiles to himself
I have been writing them down ever since.

Young Woman, Breathless Discovery

English. The moment something impossible becomes real.

Show prompt
Voice Female, early 20s. Bright soprano. Slightly breathy. American West Coast. The kind of voice that smiles while speaking. Breathless awe, tumbling over words. Scene An open field, something glowing in front of her
She freezes, eyes wide
Oh my god. Oh my god, it is real. I thought they were lying, I thought it was just some internet thing but it is actually here and it is glowing and I do not know what to do with my hands right now.

Grand-mère (French Grandmother)

French. Warm like wool blankets. Unhurried bedtime storytelling. (So then, little one. Shall I tell you the story of the fox who outsmarted the moon? It is a lovely story. Your mother used to ask for it every night, though today she would deny it.)

Show prompt
Voice Female, mid 70s. Soft alto, slightly reedy. Native French speaker, Parisian accent. Warm like wool blankets. Unhurried. A voice that has told a thousand bedtime stories. Scene Cozy bedroom, lamplight, child in bed Language FR
Elle s'assied au bord du lit et lisse les couvertures
Alors, mon petit. Tu veux que je te raconte l'histoire du renard qui a trompé la lune? C'est une belle histoire. Ta mère me la demandait tous les soirs, même si aujourd'hui elle le nierait.

Niño Maravillado (Spanish Child)

Spanish. Six years old. Seeing a rainbow for the first time. (Daddy! Daddy look! There are colors in the sky! Is that a rainbow? It is so big! Can we touch it? Can we go there?)

Show prompt
Voice Child, 6 years old. High-pitched, clear. Pure unfiltered wonder. Excited bursts with pauses of awe. Native Spanish speaker, Castilian accent. Scene Backyard after a rainstorm Language ES
Tira de la manga de su padre, señalando al cielo
Papa! Papa mira! Hay colores en el cielo! Es un arco iris? Es tan grande! Podemos tocarlo? Podemos ir alli?

हिंदी प्रोफेसर (Hindi Professor)

Hindi. Authority and warmth. A voice that commands attention without raising volume. (When I was young, my grandfather told me that a person is not recognized by their wealth, but by their work. I carried this across three continents. And everywhere I found it to be true.)

Show prompt
Voice Male, early 50s. Rich baritone. Native Hindi speaker. Confident, unhurried. A professor or community leader. Warmth underneath authority. Clear diction. Scene University lecture hall Language HI
He pauses at the podium and looks out at the audience
जब मैं छोटा था, मेरे दादाजी ने मुझे बताया था कि इंसान की पहचान उसकी दौलत से नहीं, बल्कि उसके काम से होती है। मैं यह बात तीन महाद्वीपों पर लेकर गया। और हर जगह यह सच पाया।

Deutsche Stimme (German Female Baritone)

German. Deep female baritone. Crisp, quiet, controlled. (You are listening to the last hour before dawn. The city is quiet now. The neon has given way to streetlight, and somewhere out there, someone is making a decision that cannot be undone.)

Show prompt
Voice Woman, mid 40s. Deep female baritone. Crisp, quiet, controlled. Classically trained voice but speaking, not singing. Precise diction. No emotion, no drama. Matter-of-fact delivery. Scene Absolute silence Language DE
She speaks calmly and clearly
Sie hören die letzte Stunde vor dem Morgengrauen. Die Stadt ist jetzt still. Das Neonlicht hat den Straßenlaternen Platz gemacht, und irgendwo da draußen trifft jemand eine Entscheidung, die sich nicht mehr rückgängig machen lässt.

Terrified Whisper

English. Hiding. Every word is a risk.

Show prompt
Voice Male, mid 30s. Whisper. Terrified. Shaking. A man hiding, trying not to be found. Every word is a risk. Breath catching between words. Scene Absolute silence
He presses against the wall, barely breathing
Listen to me. Do not turn around. The man in the grey coat has been following us since the bridge. I need you to walk to the cafe on the corner, order something, and leave through the back. I will find you. Do you understand? Nod if you understand.

British Woman, Dry Wit

English. Sardonic observation delivered with surgical precision.

Show prompt
Voice Woman, mid 40s. Strong Irish accent, Dublin. Dry, sardonic, cutting. Bone-dry wit. She sounds like she has seen it all and finds most of it beneath her. Scene Absolute silence
She speaks flatly, unimpressed
Apparently the committee has decided that what this building really needs is another meeting room. Because the problem with this organization was never the decisions. It was that we did not have enough places to avoid making them.

Emotional Acting

Emotional state shifts within a single generation. Action tags function as stage directions, controlling pacing, intensity, and delivery at specific points in the script.

Rage to Vulnerability

The moment anger breaks and something raw comes through. Hands shaking, jaw tight, eyes wet.

Show prompt
Voice A man on the edge. Explosive rage building with every sentence. Gravelly, intimidating. Italian-American inflection. Controlled fury that could snap at any moment. The kind of anger that comes from deep disrespect. Scene A dimly lit office, late at night
He stands up slowly, voice dangerously low
You come into my house, you eat my food, and then you got the nerve to tell me how to run my business. You know what your problem is? You got no respect. None. Zero.
Voice rising, finger pointing
I built this thing from nothing, nothing, while you were sitting on your ass doing God knows what. So don't come in here with that attitude. You understand me?

British GPU Rant (3-Chunk Chained)

Escalating frustration across three chained chunks. Calm to furious to defeated.

Show prompt
Voice Exasperated British male voice, educated accent, building frustration. Scene A cluttered desk surrounded by computer parts
He starts calmly but tension builds
Right, so let me get this straight. I need twenty four gigabytes of RAM just to run the text encoder. Twenty four! That's the entire bloody graphics card just for one component.
He takes a sharp breath and his voice rises
And they want four hundred quid for a card that can barely fit the model. Four hundred pounds!
He speaks rapidly with bitter sarcasm
Oh but don't worry, they say. Just use quantization, they say. Wonderful idea! Except every single quantization method turns your output into absolute rubbish. I've tried four bit, eight bit, FP eight.
He laughs bitterly
All garbage. The error compounds through forty eight layers until your cosine similarity is point one six. Might as well flip a coin.
He slumps into his chair, exhausted
So here I am. Renting GPUs by the hour from some bloke in a data center, praying the instance doesn't terminate mid-generation. Meanwhile Jensen is on stage wearing his leather jacket telling everyone AI is the future. Yeah mate, the future for people who can afford a five thousand pound graphics card.
He mutters bitterly
Absolute state of this industry.

British Woman, East London Rage

Furious. East London accent. The kind of anger that makes you check if the doors are locked.

Show prompt
Voice Shrill angry British female voice, East London accent. Screaming and furious. Scene A messy flat, pointing at the camera
She points at the camera, face twisted with rage
Are you having a bloody laugh? You absolute muppet! I told you THREE times to sort the bins out and what do I come home to? This! This absolute disaster! I swear to God if you don't get your shit together by tomorrow I am DONE. Finished! Pack your bags and piss off back to your mum's! I am NOT joking!

Eulogy (Aeschylus)

Somber. Slow. Each word lands like a stone dropped into still water.

Show prompt
Voice Woman, mid 60s. Deep. Extremely slow. Heavy with grief. Each word lands like a stone dropped into still water. Long pauses between phrases. Barely above a whisper. Scene Absolute silence
She speaks so slowly that each phrase feels like its own sentence. Heavy pauses. The weight of loss in every breath.
Even in our sleep. Pain which cannot forget. Falls drop by drop upon the heart. Until in our own despair. Against our will. Comes wisdom. Through the awful grace of God.

Terror

Sobbing. Choking on tears. Begging for help.

Show prompt
Voice Woman, late 20s. Voice shaking violently. Hyperventilating. Sobbing. Choking on tears. Words barely coming out between gasps for air. Throat tight with panic. Speaking through crying. Scene Absolute silence
She gasps for air between sobs, voice breaking on every word, barely able to speak through the tears
Please. Please help me. I can hear them downstairs. They broke the window. My baby is with me. Please send help. Please hurry. Please.

Villain

The laugh comes first. Then the cold precision.

Show prompt
Voice Male. Deep, resonant, theatrical voice dripping with contempt and dark amusement. Dramatic pauses. Shifting between sinister whispers and booming declarations. Scene Absolute silence
He laughs, quiet at first, then louder, then speaks with cold precision
Heheheh. Hahahaha! Oh I have waited so long for this. They told me you were clever. They said be careful. And here you are, on your knees, with nothing left. Tell me. Was it worth it? All that running?

Scene-Aware Audio

Voice and environment are generated together in the same pass. Rain, thunder, crowds, sirens. Not layered in post.

Rain and Thunder

A fisherman shouting over a storm. Rain, wind, and thunder are generated alongside the voice in the same diffusion pass.

Show prompt
Voice Male, mid 40s. Baritone. Weathered. Urgent, projecting over wind and rain. Scene Open dock in a thunderstorm, heavy rain, waves crashing against the pier
Heavy rain and wind howling
He cups his hands and shouts over the wind and rain
Get the lines! Get the lines now! She is pulling loose! If we lose this boat we lose everything!
Thunder cracks overhead
He screams louder
Move! I said move!

Radio Drama

A tense phone call in a dark room. Phone ringing, rain on the window, and thunder generated as part of the scene.

Show prompt
Voice Tense male whisper Scene Dark room, heavy rain pattering against a window
A phone rings twice then stops
He picks up the receiver with a click and speaks in a low tense whisper
Its done. The package is at the location we discussed.
Thunder rumbles in the distance
He continues urgently
You have exactly thirty minutes before they sweep the building. Do not be late.

Bank Heist

Spanish. An armed robbery in progress. Alarm bells, sirens, and screaming generated alongside the speech.

Show prompt
Voice Female, early 30s. Castilian Spanish. Sharp, commanding, ice cold under pressure. Scene Inside a bank during an armed robbery, alarm bells ringing, police sirens wailing outside, people screaming Language ES
Alarm bells ringing loudly, people screaming and crying on the floor
She screams at the bank teller over the blaring alarm
Abre la caja! Abre la caja ahora! No tengo todo el dia!
Police sirens growing louder outside, banging on the front door
She grabs the phone and speaks urgently, her voice dropping to controlled and cold
Ya tengo el dinero. Trae la furgoneta. Dos minutos. Si no llegas te dejo aqui.

Cooking Show

Italian home cook. Oil sizzling, pots bubbling, garlic hitting a hot pan. All generated in the same pass as the voice.

Show prompt
Voice Female, mid 30s. Warm, enthusiastic. Italian accent. A home cook who treats every meal like a celebration. Scene Busy home kitchen, oil sizzling in a hot pan, pots bubbling on the stove
Oil sizzling loudly in a hot pan, a pot bubbling on the stove
She talks over the sizzling, gesturing with a wooden spoon, energetic and happy
Okay now this is the important part. You wait until the oil is really hot, you see the smoke? That is when you drop the garlic in.
Garlic hits the hot oil with a loud sizzle and crackle
She stirs quickly, laughing
Beautiful! You smell that? Now we add the tomatoes and let it all come together.

Kids’ Voices

Natural children’s voices with age-appropriate mannerisms, not pitch-shifted adults.

Kid Explaining Dinosaurs

A tiny professor delivering facts with absolute authority.

Show prompt
Voice Boy, 8 years old. Small clear voice. Speaking carefully like he is the authority on this subject. A child explaining something important to someone younger. Scene Absolute silence
He speaks seriously, like a tiny professor
Okay so dinosaurs. They were really really big, like bigger than this whole house. And they lived a million billion years ago. And you know what happened? A giant rock came from space and hit the earth and then it got really cold and they all had to go away. But birds are actually dinosaurs. So technically we have dinosaurs right now.

Recess Report

Too much happened today. Not enough time to tell it all.

Show prompt
Voice Boy, 6 years old. High-pitched child voice. Excited, talking fast, jumping between topics. A kid who has too much to say and not enough time. Scene Absolute silence
He talks with his mouth half full
And then at recess Marcus said he could do a backflip but he totally cannot because I saw him try and he just fell on his back. And then we found a really big beetle near the fence and it had like a hundred legs. Well not a hundred but a lot. And oh! I forgot to tell you we have a spelling test tomorrow but I already know all the words.

Bedtime Negotiation

Building a legal case against an unjust eight-thirty bedtime.

Show prompt
Voice Girl, 7 years old. High-pitched child voice. Indignant. Building a logical argument. Scene Absolute silence
She crosses her arms
But that is not fair because yesterday you said I could stay up until nine and now you are saying eight thirty. And Jake gets to stay up way later than me and he is only one year older.

Long-form Narration

Built for audiobooks, podcasts, and documentary narration. Consistent voice across any length, no upper limit on duration.

Every sample below is a raw, unedited, single-shot generation.

Pyramus and Thisbe (3.5 minutes)

English, female voice. A full retelling across 18 chained chunks with shifting emotional registers.

Show prompt
Voice Female, mid 40s. Rich mezzo-soprano. British stage actress. Theatrical but not overwrought. Clear diction, emotional range, sustained performance energy. Scene Absolute silence
Dramatic opening, setting the ancient world
In Babylon, where ancient walls rose high against the desert sun, there lived two lovers separated by stone and silence. Pyramus, beautiful as the dawn, and Thisbe, whose voice could make the jasmine bloom. Their families shared a wall but shared nothing else. No greeting. No kindness. Only cold stone between two houses that refused to speak.
Voice softens with tenderness, intimate and warm
But the wall had a crack. Thin as a hair, invisible to anyone not looking for it. And through that crack they pressed their lips and whispered. Words too soft for anyone else to hear. I love you. I miss you. I cannot bear another day like this. Love finds a way, they said. Love always finds a way.
Quickening pace, conspiratorial excitement building
And so they made their plan. The old mulberry tree beyond the city walls, where the white berries hung like pearls in the moonlight. They would meet there at nightfall. No more whispers through stone. No more invisible kisses pressed against cold plaster. At last they would be whole.
Voice darkens, slower, ominous weight settling in
Thisbe arrived first. The moon was high and the tree was silver with it. But beneath its branches something moved. A lioness. Fresh from a kill, her jaws still wet with blood. Thisbe ran. She ran without thinking, without breathing, and as she fled her veil slipped from her shoulders and fell to the ground. The lioness, curious, mouthed the silk. Tore it. Left it stained and ruined in the dirt.
Devastation. Raw. Voice breaking.
When Pyramus found that veil, bloodied and torn beneath the tree, he did not wait. He did not call her name. He did not think. He drew his sword and drove it beneath his ribs and fell upon the roots of the tree. His blood ran dark into the earth. And the white berries above him turned red. One by one. Like tears.
Almost a whisper. Grief beyond words.
And when Thisbe came back. When she found him there, still warm, still breathing his last breath against the ground. She did not scream. She did not run. She lay beside him. She took the sword. And she followed him into the dark without a single word.
Quiet resolution, elegiac, a closing benediction
The mulberry fruit, once white, has been dark ever since. Stained with the blood of two people who loved too much and waited too long. The gods, for once, showed mercy. They let their ashes rest together in a single urn. What the wall kept apart in life, death made whole. And the fruit remembers. It has remembered for three thousand years.

Bhagavad Gita, Chapter 2 (1.5 minutes)

Hindi, male voice. Krishna speaks to Arjuna about the eternal soul. (You grieve for those who should not be grieved for, and yet you speak words of wisdom. The wise grieve neither for the living nor for the dead...)

Show prompt
Voice Male, mid 60s. Deep baritone. Native Hindi speaker. The voice of divine authority speaking with infinite patience. Unhurried, each word carries the weight of eternity. A guru delivering truth. Scene Absolute silence Language HI
He speaks with calm authority, as if addressing all of humanity
तुम उनके लिए शोक करते हो जो शोक के योग्य नहीं हैं, और फिर भी ज्ञान की बातें करते हो। जो बुद्धिमान हैं वे न जीवित के लिए और न मृत के लिए शोक करते हैं।
His voice deepens, revealing an eternal truth
ऐसा कभी नहीं हुआ कि मैं नहीं था, या तुम नहीं थे, या ये सब राजा नहीं थे। और ऐसा भी नहीं होगा कि इसके बाद हम नहीं रहेंगे...
Quiet finality, absolute certainty
जैसे मनुष्य पुराने वस्त्र त्यागकर नये वस्त्र धारण करता है, वैसे ही आत्मा पुराने शरीर को त्यागकर नया शरीर धारण करती है। इसे शस्त्र काट नहीं सकते, अग्नि जला नहीं सकती, जल भिगो नहीं सकता, और वायु सुखा नहीं सकती।

Cien Años de Soledad (3 minutes)

Spanish, male voice. The opening pages of One Hundred Years of Solitude. (Many years later, as he faced the firing squad, Colonel Aureliano Buendia was to remember that distant afternoon when his father took him to discover ice...)

Show prompt
Voice Male, mid 50s. Rich, warm baritone. Native Spanish speaker, Colombian. Audiobook narrator. Extremely slow, deliberate pacing. Each word savored. Scene Absolute silence Language ES
He speaks slowly, reverently, as if reading scripture
Muchos anos despues, frente al peloton de fusilamiento, el coronel Aurreliano Buendia habia de recordar aquella tarde remota en que su padre lo llevo a conocer el hielo. Macondo era entonces una aldea de veinte casas de barro y canabrava, construidas a la orilla de un rio de aguas diafanas que se precipitaban por un lecho de piedras pulidas, blancas y enormes como huevos prehistoricos.
His voice warms, slower still, painting the world with each phrase
El mundo era tan reciente, que muchas cosas carecian de nombre, y para mencionarlas habia que senalarlas con el dedo. Todos los anos, por el mes de marzo, una familia de gitanos desarrapados plantaba su carpa cerca de la aldea, y con un grande alboroto de pitos y timbales daban a conocer los nuevos inventos.
Wonder creeping in, but still measured and unhurried
Primero llevaron el iman. Un gitano corpulento, de barba montaraz y manos de gorrion, que se presento con el nombre de Melquiades, hizo una truculenta demostracion publica de lo que el mismo llamaba la octava maravilla de los sabios alquimistas de Macedonia. Fue de casa en casa arrastrando dos lingotes metalicos, y todo el mundo se espanto al ver que los calderos, las sartenes, las tenazas y los anafes se caian de su sitio.
Voice deepens, the obsession taking root
Jose Arcadio Buendia, cuya desaforada imaginacion iba siempre mas lejos que el ingenio de la naturaleza, y aun mas alla del milagro y la magia, penso que era posible servirse de aquella invencion inutil para desentranar el oro de la tierra. Melquiades, que era un hombre honrado, le previno. Para eso no sirve. Pero Jose Arcadio Buendia no creia en aquel tiempo en la honradez de los gitanos.
Reflective, the futility settling in
Asi que cambio su mula y una partida de chivos por los dos lingotes imantados, y recorrio la region, palmo a palmo, incluso el fondo del rio, arrastrando los dos lingotes de hierro y recitando en voz alta el conjuro de Melquiades. Lo unico que logro desenterrar fue una armadura del siglo quince con todas sus partes soldadas por un cascote de oxido, cuyo interior tenia la resonancia hueca de un enorme calabazo lleno de piedras.
Quiet amusement at the discovery
Cuando Jose Arcadio Buendia y los cuatro hombres de su expedicion lograron desarticular la armadura, encontraron dentro un esqueleto calcificado que llevaba colgado del cuello un relicario de cobre con un rizo de mujer.

Die Verwandlung (1 minute)

German, female voice. The opening of Kafka's Metamorphosis. (When Gregor Samsa woke one morning from troubled dreams, he found himself transformed in his bed into a monstrous vermin...)

Show prompt
Voice Woman, mid 40s. Deep female baritone. Crisp, precise diction. Native German speaker, Austrian-inflected Hochdeutsch. Clinical narration that slowly reveals its horror. The voice of someone describing the unthinkable as if it were entirely ordinary. Scene Absolute silence Language DE
She begins matter-of-factly, as if reading a report
Als Gregor Samsa eines Morgens aus unruhigen Traeumen erwachte, fand er sich in seinem Bett zu einem ungeheueren Ungeziefer verwandelt. Er lag auf seinem panzerartig harten Ruecken und sah, wenn er den Kopf ein wenig hob, seinen gewoelbten, braunen, von bogenfoermigen Versteifungen geteilten Bauch, auf dessen Hoehe sich die Bettdecke, zum gaenzlichen Niedergleiten bereit, kaum noch erhalten konnte.
A note of unease creeps into the clinical tone
Seine vielen, im Vergleich zu seinem sonstigen Umfang klaeglich duennen Beine flimmerten ihm hilflos vor den Augen. Was ist mit mir geschehen, dachte er. Es war kein Traum.
She pauses, then continues with growing resignation
Sein Zimmer, ein richtiges, nur etwas zu kleines Menschenzimmer, lag ruhig zwischen den vier wohlbekannten Waenden. Ueber dem Tisch, auf dem eine auseinandergepackte Musterkollektion von Tuchwaren ausgebreitet war, hing das Bild, das er vor kurzem aus einer illustrierten Zeitschrift ausgeschnitten und in einem huebschen, vergoldeten Rahmen untergebracht hatte.

Alice in Wonderland, Chapter 1 (15 minutes)

English, female voice. The complete first chapter of Alice's Adventures in Wonderland. Generated in 9 minutes (1.75x faster than real-time) on a single RTX A6000. 95.6% average word accuracy across all chunks, raw and unedited.

Show prompt
Voice Young woman, early 20s. British. Bright, clear soprano. Curious and slightly bored. A well-read girl with an active imagination. Scene Riverbank on a lazy summer afternoon. Gentle water flowing, birds singing, bees buzzing in the grass.
She sits on the riverbank next to her sister, restless, picking at the grass, narrating with languid boredom
Alice was beginning to get very tired of sitting by her sister on the bank, and of having nothing to do. Once or twice she had peeped into the book her sister was reading, but it had no pictures or conversations in it. And what is the use of a book, thought Alice, without pictures or conversation?...
She sits bolt upright, eyes wide with surprise, her voice quickening with curiosity
There was nothing so very remarkable in that. Nor did Alice think it so very much out of the way to hear the Rabbit say to itself, Oh dear! Oh dear! I shall be late!...
Her voice shifts to wonder and slight unease as she describes the descent
The rabbit-hole went straight on like a tunnel for some way, and then dipped suddenly down...
Her voice grows dreamy and meandering, drifting between thoughts as sleep creeps in
Down, down, down. There was nothing else to do, so Alice soon began talking again...
She gasps with wonder, her voice filling with longing as she describes the garden
Alice opened the door and found that it led into a small passage, not much larger than a rat-hole. She knelt down and looked along the passage into the loveliest garden you ever saw...
Her voice breaks with frustration, on the verge of tears
But, alas for poor Alice! When she got to the door, she found she had forgotten the little golden key...
She speaks with renewed determination, a practical resolve in her voice
Well, I will eat it, said Alice, and if it makes me grow larger, I can reach the key. And if it makes me grow smaller, I can creep under the door. So either way I will get into the garden, and I do not care which happens!

How it Works

The pipeline has four stages: text encoding (Gemma 3 12B, bf16), audio diffusion (8-step denoising), post-processing (vocal isolation, validation, silence trimming), and optional voice identity transfer via SeedVC.

A 20-second clip takes about 5-8 seconds end-to-end on a RTX 4090. Minimum hardware is 16GB VRAM with CPU streaming for the text encoder. Standard all-on-GPU configuration requires 24GB.

Pyramus and Thisbe

To show what the full pipeline looks like end to end, we produced a short film of Pyramus and Thisbe, one of the oldest tragic love stories in Western literature. Ovid first told it in his Metamorphoses, and Shakespeare later drew from it for both Romeo and Juliet and the play within a play in A Midsummer Night’s Dream. Every scene was directed and generated from a single script using Scenema. Narration was produced with Scenema Audio. The exact audio output is demonstrated above in the long narration samples.


Scenema Audio is fully open source under the MIT license. The inference code and all audio processing code are on GitHub and can be run as a Docker service. The model weights are on HuggingFace. Scenema is free to get started. You can begin generating voiceovers with Scenema Audio inside the platform today.