Fish Audio: Drama 3
Drama 3 turns a written script into directed speech. Shape emotion and delivery with plain-language cues, add pauses and sounds, guide pronunciation and build multi-speaker scenes. Its published speech workflow supports more than 80 languages, including language changes within a sentence.
About the model
This description covers the model’s capabilities. Available in Ailoot inputs, outputs and options are listed for the selected method below.
Drama 3 is Fish Audio’s text-to-speech model for a performed script. It produces speech audio from the words you supply and gives the scriptwriter control over delivery. A line can carry an emotional direction, a local change in voice performance or a pronunciation instruction, so the input contains both the spoken content and the way it should sound.
Write the words and direct the performance
Prepare the exact words the listener should hear before adding performance cues. Keep important names, dates and instructions clear in the script, and use punctuation to establish the intended sentence rhythm. Then place a plain-language direction in square brackets before the line it concerns. Describe the desired emotion or delivery in concrete terms, such as a restrained announcement, a tense exchange or a warm explanation.
Drama 3 also supports local controls within a line. Inline instructions can introduce sounds and pauses, while paired delivery tags can affect particular words. Keep the scope of each instruction clear: a direction for one phrase should not accidentally become the direction for the entire scene. Read the script once for meaning and once for performance before synthesis.
Scenes, languages and pronunciation
The published model supports multiple speakers within one request. Organise dialogue into clear speaker turns and give each turn a purpose: a question, an interruption, an explanation or a response. For a lesson or an explainer, the same approach can separate a narrator from a quoted character without rewriting the material as one undifferentiated monologue.
Drama 3 supports more than 80 languages and can change language within a sentence while keeping the voice. Prepare multilingual passages in their intended spoken languages, rather than assuming that an English instruction supplies a translation. Decide how foreign names and borrowed terms should sound, and use the model’s phoneme controls when their pronunciation needs precise direction.
Useful audio artifacts
The model fits character dialogue, narrated lessons, short explainers, product announcements and scripted voice responses. Keep content and acting choices together: an educational passage needs intelligible terminology and purposeful pauses; a character scene needs clearly separated turns and a consistent dramatic intention. Those are writing decisions that can be reviewed before any audio is generated.
After synthesis, compare the spoken words with the finished script. Listen for omitted phrases, misplaced pauses, unclear speaker changes and the pronunciation of protected names. Retain the script and its delivery cues with the audio so a correction can target the affected line. Keep the final spoken content and its line-level performance choices together when preparing a revision.
Capabilities
- Text-to-speech performance from a finished script.
- Plain-language square-bracket directions before a line.
- Inline sounds and pauses, with word-local paired delivery tags.
- Multi-speaker scenes and more than 80 spoken languages.
- Phoneme controls for names and specialist terminology.
Use cases
- Character dialogue — A script with clearly separated speaker turns and performance directions. Result: A spoken scene with deliberate changes of delivery.
- Narrated explanation — A finished lesson with terminology and pauses marked. Result: Speech audio that can be reviewed against the lesson text.
- Multilingual announcement — Final text in the intended languages with pronunciation choices retained. Result: A spoken passage with controlled language transitions.
- Scripted voice response — A concise response with an explicit tone and the words to be spoken. Result: A reusable speech segment for the surrounding application.
Limits
- Input and result — The published model takes text and returns speech audio.
- Control scope — Attach a direction to its intended line or phrase, and keep paired delivery tags properly scoped.
- Review unit — Retain the exact script and cues with the audio; review pronunciation and speaker turns at the line level.
Ailoot Ultra · Speech POST /v1/audio/speech fish-audio/drama-3
fish-audio/drama-3Overview
Drama 3 turns a written script into directed speech. Shape emotion and delivery with plain-language cues, add pauses and sounds, guide pronunciation and build multi-speaker scenes. Its published speech workflow supports more than 80 languages, including language changes within a sentence.
- Accepted inputs
- Text
- Results
- Audio
Available in Ailoot · inputs & methods
Ailoot Ultra · Speech POST /v1/audio/speech
- Input types
text- Output types
audio
| Field | Type | Requirement | Values & limits | Default |
|---|---|---|---|---|
Fieldmodel | Typestring | RequirementRequired | Values & limits fish-audio/drama-3 | Default— |
Fieldinput | Typestring | RequirementRequired | Values & limits Minimum length: 1 | Default— |
Fieldvoice | Typestring | RequirementRequired | Values & limits Minimum length: 1 | Default— |
Fieldresponse_format | Typestring | RequirementOptional | Values & limits mp3, pcm | Defaultpcm |
Fieldprovider | Typeobject | RequirementOptional | Values & limits | Default— |
Method parameters and lifecycle
| Parameter | Contract |
|---|---|
Parametercontent_type | Contractapplication/json |
Parameterrequired | Contractmodel, input, voice |
Parametermaterial_types | Contracttext |
Parametermaterial_conditions | ContractSelect an actual supported voice ID; no literal default/voice has been invented for this model. |
Lifecycle: synchronous_result_then_owned_library
GET /v1/media/tasks/{task_id} · actual returned task_id or media library record
GET /v1/media/assets/{asset_id}/content
GET /v1/media/assets
GET /v1/media/assets/{asset_id}
Pricing
Tariff references do not add request options. Supported parameters are listed for the selected method.
- Fast mode
- Not specified When fast mode is selected
- Flex mode
- Not specified When flex mode is selected
- Maximum reasoning effort
- Not specified When maximum reasoning effort uses the default adjustment
Ailoot rates
Prices by configuration
Match each charge to the conditions shown. Use one rate for each set of alternative configurations.
| Configuration | Rate schedule | Billing unit | Charge |
|---|---|---|---|
| ConfigurationBase token rates | Rate scheduleBase rate | Billing unit— | Charge Input · Input tokens — / 1K tokens Output · Output tokens — / 1K tokens Input · Cache read — / 1K tokens Input · Cache write — / 1K tokens Input · Cache write, 1 hour — / 1K tokens |
| Configurationinput UTF-8 byte | Rate schedule— | Billing unitper input UTF-8 byte | Charge Input · input UTF-8 byte $0.00002 per input UTF-8 byte |
Ailoot reserves an estimated amount before a request. After the result, it settles the final charge from measured usage. Streaming settlement uses the final usage record. How reservation and settlement work →
Example requests
Ailoot Ultra · Speech
set -eu
: "${AILOOT_API_KEY:?Set your Ailoot API key}"
: "${AILOOT_INPUT_TEXT:?Set the actual documented input value}"
export AILOOT_INPUT_TEXT
: "${AILOOT_VOICE:?Set the actual documented input value}"
export AILOOT_VOICE
python3 - <<'PY_REQUEST'
import json,os
from pathlib import Path
request={'model': 'fish-audio/drama-3', 'input': '${ACTUAL_INPUT_TEXT}', 'voice': '${ACTUAL_SUPPORTED_VOICE}'}
request['input']=os.environ['AILOOT_INPUT_TEXT']
request['voice']=os.environ['AILOOT_VOICE']
Path('request.json').write_text(json.dumps(request,ensure_ascii=False),encoding='utf-8')
PY_REQUEST
curl --fail-with-body --silent --show-error https://api.ailoot.ai/v1/audio/speech \
-H "Authorization: Bearer $AILOOT_API_KEY" -H 'Content-Type: application/json' \
--data-binary @request.json -D response-headers.txt -o response.bin
# Retain the complete response and headers, every actual output, original returned identifiers and all usage fields. Recover the original operation without submitting creation again.
Receive and save results
Ailoot Ultra · Speech
Complete binary audio or original audio_url response, headers and every owned asset.
GET /v1/media/tasks/{task_id}GET /v1/media/assets/{asset_id}/contentGET /v1/media/assetsGET /v1/media/assets/{asset_id}
Other content reads
set -eu
: "${AILOOT_API_KEY:?Set your Ailoot API key}"
curl --fail-with-body --silent --show-error https://api.ailoot.ai/v1/media/assets \
-H "Authorization: Bearer $AILOOT_API_KEY" -o owned-assets.json
# Preserve every returned asset and follow its exact signed content URL. Do not construct a signed token.
Keep all completed results
Keep every complete returned response, file and content part.
Retain the complete response and headers, every actual output, original returned identifiers and all usage fields. Recover the original operation without submitting creation again.
If the original response needs checking
Keep the returned X-Client-Request-ID for reconciliation. Read only actual returned task_id/asset_id values or the owned media library; never build a task URL from the operation ID and never resubmit creation.
Aliases
Model aliases
Ailoot Ultra fish-audio/drama-3
Method path aliases
Use the method paths shown above.
Related caller IDs
FAQ · Fish Audio: Drama 3
- What inputs and results are available?
POST /v1/audio/speech·fish-audio/drama-3:text→audio- Which fields does the request use?
POST /v1/audio/speech·fish-audio/drama-3:model,input,voice,response_format,provider