GET

/v1/schemas/openrouter/audio/audio/speech

Accept:

kind inputactivity audiohash 4c34e854…extracted Nov 13, 2026version currentsource upstream spec ↗

field	recorded source
schema	Source linked above. upstream-spec Source trace `0d19a8453874651361f64fbd51045d9d1d4868b39f7eb85b9866fee302c50de2` { "fetchedAt": 1794547804, "extractorVersion": "2", "verifiedAt": null }

Request body

11 properties · 2 required · 11 $defs

property	type	description	constraints
input*	string \| SpeechTurn[]	Text to synthesize, or a list of turns for multi-speaker input. Each turn has its own text, voice, and instructions. Multi-speaker input is supported by Gemini TTS models, by Fish Audio models other than S1, and by Speechify dialogue models.
model*	string	TTS model identifier
input_references	SpeechInputReference[]	Reference content for stateless voice cloning or voice design. Audio mode: one to three `input_audio` parts, each optionally paired with a `text` part carrying its transcript (a single clip accepts its transcript before or after it; with multiple clips each tr…
instructions	string	Delivery instructions for the whole request, such as tone, pacing, or emotion. Supported by OpenAI gpt-4o-mini-tts, Gemini TTS, and Fish Audio models. Ignored by other providers.
provider	object	Provider configuration: provider routing preferences (`order`, `only`, `ignore`, `allow_fallbacks`), data policy routing preferences (`zdr`, `data_collection`) and provider-specific passthrough options
response_format	enum	Audio output format	mp3 \| pcm · default: "pcm"
session_id	string	A unique identifier for grouping related requests (e.g., a conversation or agent workflow). Used for observability grouping in Broadcast and private logging; never sent to the provider. If provided in both the request body and the x-session-id header, the body…	length ≤ 256
speed	number	Playback speed multiplier. Honored by models that support it (e.g. OpenAI TTS). Other providers either ignore it or return a 400 for a non-default value when the model has no speed control.	format: double
trace	TraceConfig	Metadata for observability and tracing. Known keys (trace_id, trace_name, span_name, generation_name, parent_span_id) have special handling. Additional keys are passed through as custom metadata to configured broadcast destinations.
user	string	A unique identifier representing your end-user. Forwarded to Broadcast and private logging as the end-user id; never sent to the provider.	length ≤ 256
voice	string	Voice identifier (provider-specific).	length ≥ 1

Types

11 $defs

ProviderNameenum

Provider-specific options keyed by provider slug. The provider segment is case-insensitive; an endpoint tag after `/` must match the endpoint slug exactly. Only options for the matched provider are forwarded; the rest are ignored. Unrecognized keys are silentl…

01ai	object
ai21	object
aion-labs	object
akashml	object
alibaba	object
amazon-bedrock	object
amazon-bedrock/claude-on-aws	object
amazon-nova	object
ambient	object
antdigital	object
anthropic	object
anthropic/2	object
anyscale	object
arcee-ai	object
assemblyai	object
atlas-cloud	object
atoma	object
avian	object
azure	object
baidu	object
baseten	object
black-forest-labs	object
byteplus	object
centml	object
cerebras	object
chutes	object
cirrascale	object
clarifai	object
claude-on-aws	object
cloudflare	object
cohere	object
coreweave	object
cosine	object
crofai	object
crucible	object
crusoe	object
darkbloom	object
databricks	object
decart	object
deepgram	object
deepinfra	object
deepseek	object
dekallm	object
digitalocean	object
elevenlabs	object
enfer	object
fake-provider	object
featherless	object
fireworks	object
fish-audio	object
friendli	object
gmicloud	object
google-ai-studio	object
google-vertex	object
gopomelo	object
groq	object
heygen	object
huggingface	object
hyperbolic	object
hyperbolic-quantized	object
inception	object
inceptron	object
inferact-vllm	object
inference-net	object
infermatic	object
inflection	object
inocloud	object
io-net	object
ionstream	object
klusterai	object
krea	object
kt	object
lambda	object
lepton	object
liquid	object
lynn	object
lynn-private	object
makora	object
mancer	object
mancer-old	object
mara	object
meta	object
minimax	object
mistral	object
modal	object
modelrun	object
modular	object
moonshotai	object
morph	object
ncompass	object
near-ai	object
nebius	object
nex-agi	object
nextbit	object
nineteen	object
novita	object
nvidia	object
octoai	object
ollama	object
open-inference	object
openai	object
parallel	object
parasail	object
pareto-alpha	object
particle-ai	object
pearl-research	object
perceptron	object
perplexity	object
phala	object
poolside	object
primalabs	object
primeintellect	object
quiver	object
recraft	object
recursal	object
recursive	object
reflection	object
reka	object
relace	object
replicate	object
respan	object
river	object
runway	object
sail-research	object
sakana	object
sakana-ai	object
sambanova	object
sambanova-cloaked	object
scaledown	object
seed	object
sf-compute	object
siliconflow	object
sourceful	object
speechify	object
stealth	object
stepfun	object
streamlake	object
switchpoint	object
targon	object
tencent	object
tenstorrent	object
thinkingmachines	object
together	object
together-lite	object
typesafe	object
ubicloud	object
unbiased	object
upstage	object
venice	object
voyageai	object
wafer	object
wandb	object
wandb-legacy	object
xai	object
xiaomi	object
z-ai	object

SpeechInputstring | SpeechTurn[]

Text to synthesize, or a list of turns for multi-speaker input. Each turn has its own text, voice, and instructions. Multi-speaker input is supported by Gemini TTS models, by Fish Audio models other than S1, and by Speechify dialogue models.

SpeechInputReferenceobject | object | object

Reference content part for stateless voice cloning or voice design

SpeechInputReferenceAudioobject

Reference audio input for stateless voice cloning. Up to three parts per request; the Nth audio part is addressable from `input` as `@AudioN` on providers that support multiple references.

input_audio*	SpeechInputReferenceAudioInput	Reference audio input object
type*	enum		input_audio

SpeechInputReferenceAudioInputobject

Reference audio input object

data	string	Base64-encoded reference audio (optionally a data URI). Supported audio formats are provider-specific. Limited to 20 MiB of base64 (15 MiB of decoded audio). Exactly one of `data` or `url` is required.	length ≥ 1 · length ≤ 20971520
format	string	Audio format of the reference audio (e.g., wav, mp3). Optional; most providers detect the format from the audio bytes.
url	string	Public http(s) URL of the reference audio. OpenRouter downloads it (15 MiB max) and forwards the bytes, never the URL. Exactly one of `data` or `url` is required.	length ≤ 2048 · format: uri

SpeechInputReferenceImageobject

Reference image. On its own it describes the desired voice (voice design); alongside an `input_audio` part it is a scene image describing the setting the cloned voice speaks in. At most one per request. Only routed to endpoints that support image references.

image_url*	SpeechInputReferenceImageInput	Reference image input object
type*	enum		image_url

SpeechInputReferenceImageInputobject

Reference image input object

url*

string

JPEG, PNG, or WebP reference image as a base64 data URI or a public http(s) URL. Remote images are downloaded (15 MiB max) and forwarded as bytes, never as the URL.

length ≥ 1 · length ≤ 20971520

SpeechInputReferenceTextobject

Transcript of an `input_audio` part

text*	string	Transcript of an `input_audio` part. With a single clip it may appear before or after the clip; with multiple clips it must immediately follow the clip it transcribes.	length ≤ 10000
type*	enum		text

SpeechTurnobject

instructions	string	Delivery instructions for this turn, such as tone, pacing, or emotion. Overrides the top-level `instructions`.
text*	string	Text to synthesize
voice	string	Voice for this turn. Defaults to the top-level `voice`.	length ≥ 1

TraceConfigobject

Metadata for observability and tracing. Known keys (trace_id, trace_name, span_name, generation_name, parent_span_id) have special handling. Additional keys are passed through as custom metadata to configured broadcast destinations.

generation_name	string
parent_span_id	string
span_name	string
trace_id	string
trace_name	string

Validate a payload

before you spend tokens

shell

$ curl -X POST https://modelschemas.com/v1/validate \
    -d '{"provider":"openrouter","endpointId":"audio/speech","payload":{…}}'
→ {"valid": false, "errors": [{"path": "#", "keyword": "required", …}]}