> ## Documentation Index > Fetch the complete documentation index at: https://docs.withhopper.com/llms.txt > Use this file to discover all available pages before exploring further. # Generate speech > Native speech synthesis with full control of container, sample rate, and encoding. Returns raw audio bytes; set stream true for progressive relay of WAV/PCM bytes. ## OpenAPI ````yaml /api-reference/openapi.json post /tts/bytes openapi: 3.1.0 info: title: Hopper API version: 1.0.0 description: >- Inference for voice agents: LLM, text to speech, and speech to text at api.withhopper.com. OpenAI-compatible routes are served under /v1; native routes are served at the root. All errors use the OpenAI envelope. servers: - url: https://api.withhopper.com security: - bearerAuth: [] tags: - name: LLM description: OpenAI-compatible chat completions and model listing. - name: Text to Speech description: >- Speech synthesis: native /tts routes and the OpenAI-compatible /v1/audio/speech. - name: Speech to Text description: Batch transcription. Streaming runs over WS /stt/websocket. - name: Voices description: Voice library, cloning, profiles, and localization. - name: Pronunciation description: Pronunciation dictionaries applied at synthesis time. - name: Jobs description: Status of async voice jobs. - name: Health description: Unauthenticated liveness check. paths: /tts/bytes: post: tags: - Text to Speech summary: Generate speech description: >- Native speech synthesis with full control of container, sample rate, and encoding. Returns raw audio bytes; set stream true for progressive relay of WAV/PCM bytes. requestBody: required: true content: application/json: schema: $ref: '#/components/schemas/TtsRequest' responses: '200': description: Audio bytes. Content-Type follows output_format.container. headers: x-hopper-request-cost-usd: $ref: '#/components/headers/XHopperRequestCostUsd' x-hopper-credits-remaining-usd: $ref: '#/components/headers/XHopperCreditsRemainingUsd' content: audio/wav: schema: type: string format: binary audio/mpeg: schema: type: string format: binary application/octet-stream: schema: type: string format: binary '400': description: >- Invalid body. Codes: invalid_json, invalid_output_format (mp3 with stream), invalid_generation_config (speed other than 1 with stream), transcript_too_long (dictionary alias expansion past 20,000 characters). content: application/json: schema: $ref: '#/components/schemas/Error' example: error: message: Streaming TTS does not support mp3; use wav or pcm/raw. type: invalid_request_error code: invalid_output_format param: null '401': $ref: '#/components/responses/Unauthorized' '402': $ref: '#/components/responses/PaymentRequired' '404': description: >- Unknown model, voice, or pronunciation dictionary. Codes: model_not_found, voice_not_found, dict_not_found. content: application/json: schema: $ref: '#/components/schemas/Error' '409': $ref: '#/components/responses/VoiceProfileMissing' '429': $ref: '#/components/responses/RateLimited' '502': description: 'Upstream failed. Codes: upstream_unavailable, upstream_error.' content: application/json: schema: $ref: '#/components/schemas/Error' '503': description: 'No healthy upstream for this model. Code: model_offline.' content: application/json: schema: $ref: '#/components/schemas/Error' '504': description: 'Upstream timed out (300 s). Code: upstream_timeout.' content: application/json: schema: $ref: '#/components/schemas/Error' components: schemas: TtsRequest: type: object required: - model_id - transcript - voice - output_format properties: model_id: type: string description: TTS model id, e.g. omnivoice. transcript: type: string minLength: 1 maxLength: 5000 description: >- Text to synthesize, 1-5000 characters. Billed per character of the submitted transcript, before dictionary expansion. voice: type: object required: - mode - id properties: mode: type: string enum: - id id: type: string description: Voice id from GET /voices. language: type: string pattern: ^[a-zA-Z]{2,3}(-[a-zA-Z0-9]{2,8})?$ description: ISO language code. output_format: type: object required: - container - sample_rate properties: container: type: string enum: - wav - mp3 - raw description: mp3 is batch-only; streaming supports wav and raw. sample_rate: type: integer enum: - 8000 - 16000 - 22050 - 24000 - 44100 - 48000 encoding: type: string enum: - pcm_f32le - pcm_s16le - pcm_mulaw - pcm_alaw default: pcm_s16le description: wav and raw only; rejected for mp3. bit_rate: type: integer enum: - 32000 - 64000 - 96000 - 128000 - 192000 - 256000 - 320000 default: 128000 description: mp3 only. generation_config: type: object properties: speed: type: number minimum: 0.6 maximum: 1.5 description: Playback speed. Streaming requires speed 1. volume: type: number minimum: 0.5 maximum: 2 pronunciation_dict_id: type: string description: Dictionary applied before synthesis; 404 dict_not_found if unknown. stream: type: boolean default: false description: >- On /tts/bytes: relay raw WAV/PCM bytes progressively as they are generated. Ignored on /tts/sse, which always streams. Error: type: object description: OpenAI error envelope, used by every error response. properties: error: type: object properties: message: type: string type: type: string code: type: string param: type: - string - 'null' headers: XHopperRequestCostUsd: description: >- Cost of this request, 6-decimal USD string (e.g. 0.000123). Absent on LLM streaming responses; the cost is unknown until the stream ends. schema: type: string XHopperCreditsRemainingUsd: description: >- Credit balance in USD, 6-decimal string. On LLM streaming responses this is the pre-request balance. schema: type: string responses: Unauthorized: description: >- Missing, malformed, or revoked API key. Codes: invalid_api_key, api_key_disabled. content: application/json: schema: $ref: '#/components/schemas/Error' example: error: message: >- Incorrect API key provided. Check your key at https://withhopper.com/dashboard. type: invalid_request_error code: invalid_api_key param: null PaymentRequired: description: Credit balance is zero. content: application/json: schema: $ref: '#/components/schemas/Error' example: error: message: You have run out of credits. Contact us to add more. type: insufficient_quota code: insufficient_quota param: null VoiceProfileMissing: description: >- The voice has no ready profile for this model. Prepare one with POST /voices/:id/profiles. Code: voice_profile_missing. content: application/json: schema: $ref: '#/components/schemas/Error' example: error: message: Voice 'X' is not prepared for model 'Y'. type: invalid_request_error code: voice_profile_missing param: null RateLimited: description: >- Over 600 requests per minute on this key. Retry after the number of seconds in Retry-After. headers: Retry-After: description: Seconds to wait before retrying. schema: type: integer content: application/json: schema: $ref: '#/components/schemas/Error' example: error: message: >- Rate limit reached: 600 requests per minute per key. Retry after 21s. type: rate_limit_error code: rate_limit_exceeded param: null securitySchemes: bearerAuth: type: http scheme: bearer description: >- API key in the Authorization header: Bearer sk_hopper_... Keys are minted in the dashboard at withhopper.com/console/keys and shown once at creation. WebSocket connections also accept ?api_key=. ````