> ## Documentation Index > Fetch the complete documentation index at: https://docs.chariot.in/llms.txt > Use this file to discover all available pages before exploring further. # Stream text to speech > Synthesize speech and stream the audio back as it is generated, for the lowest possible time-to-first-byte over HTTP. The response is a chunked stream of raw 16-bit little-endian PCM (`audio/L16`) at 44.1 kHz, mono. Metadata is returned in response headers before the body streams: `X-inference-id`, `X-audio-url`, and `Credit-Utilized`. Because the HTTP status line is sent before generation begins, a mid-stream failure ends the stream early rather than changing the status code — always verify you received a complete stream. **Cost:** 1 credit per input character, charged up front. If generation fails, the credits are refunded. ## OpenAPI ````yaml POST /v1/tts/stream openapi: 3.1.0 info: title: Chariot AI API version: '1' description: >- Chariot AI is a text-to-speech (TTS) platform built for Indian and English voices. Convert text into natural-sounding speech over a simple REST call, an HTTP audio stream, or a low-latency WebSocket. This reference documents the public, API-key-authenticated endpoints. Speech-to-text (ASR) is coming soon. contact: name: Chariot AI Support email: support@chariot.in url: https://chariot.in servers: - url: https://api.chariot.in description: Production - url: https://dev-api.chariot.in description: Development / sandbox security: - APIKeyHeader: [] tags: - name: Text to Speech description: Convert text into speech synchronously or as a stream. - name: Voices description: List the voice catalog and manage your cloned (custom) voices. - name: Account description: Inspect your credit balance and usage. paths: /v1/tts/stream: post: tags: - Text to Speech summary: Stream text to speech description: >- Synthesize speech and stream the audio back as it is generated, for the lowest possible time-to-first-byte over HTTP. The response is a chunked stream of raw 16-bit little-endian PCM (`audio/L16`) at 44.1 kHz, mono. Metadata is returned in response headers before the body streams: `X-inference-id`, `X-audio-url`, and `Credit-Utilized`. Because the HTTP status line is sent before generation begins, a mid-stream failure ends the stream early rather than changing the status code — always verify you received a complete stream. **Cost:** 1 credit per input character, charged up front. If generation fails, the credits are refunded. operationId: streamSpeech requestBody: required: true content: application/json: schema: $ref: '#/components/schemas/TTSPayload' responses: '200': description: A chunked stream of raw PCM audio (16-bit LE, 44100 Hz, mono). headers: X-inference-id: description: Unique id for this generation. schema: type: string format: uuid X-audio-url: description: >- Persistent URL to the fully assembled WAV once the stream completes. schema: type: string format: uri Credit-Utilized: description: Credits charged for this request. schema: type: integer content: audio/L16: schema: type: string format: binary '401': $ref: '#/components/responses/Unauthorized' '403': $ref: '#/components/responses/InsufficientCreditsOrPlan' '404': $ref: '#/components/responses/VoiceNotFound' '422': $ref: '#/components/responses/ValidationError' '500': $ref: '#/components/responses/ServerError' components: schemas: TTSPayload: type: object required: - voice_id - text properties: voice_id: type: string format: uuid description: The voice to synthesize with. Get valid ids from `GET /v1/voices`. example: bac7d666-094d-4698-91fa-741d60fce662 text: type: string description: >- The text to synthesize. 1 to 500 characters after trimming whitespace. minLength: 1 maxLength: 500 example: Hello from Chariot AI. model_type: $ref: '#/components/schemas/TTSModelType' TTSModelType: type: string description: The TTS model to use. Currently only `v0` is available. enum: - v0 default: v0 Error: type: object description: Standard error envelope returned by all endpoints. properties: statusCode: type: integer description: HTTP status code, echoed in the body. error: type: string description: Machine-readable error class name. message: type: string description: Human-readable description of what went wrong. required: - statusCode - error - message responses: Unauthorized: description: >- Authentication failed — the API key is missing, invalid, expired, or the request origin is not allowed for this key. content: application/json: schema: $ref: '#/components/schemas/Error' examples: missing: summary: No credentials provided value: statusCode: 401 error: UnauthorizedException message: No API key or user credentials provided invalid: summary: Invalid or unknown API key value: statusCode: 401 error: UnauthorizedException message: Invalid API Key or user credentials InsufficientCreditsOrPlan: description: >- The request is authenticated but not permitted: the workspace has insufficient credits, or the plan does not include this product. content: application/json: schema: $ref: '#/components/schemas/Error' examples: insufficient_credits: summary: Not enough credits value: statusCode: 403 error: InsufficientCreditsException message: Workspace has 4 credits; needs 128 not_on_plan: summary: Product not available on plan value: statusCode: 403 error: HTTPException message: >- Product model tts.standard is not available on the current plan VoiceNotFound: description: >- The requested voice does not exist or is not accessible to your workspace. content: application/json: schema: $ref: '#/components/schemas/Error' example: statusCode: 404 error: EntityNotFoundException message: >- Voice with id: 00000000-0000-0000-0000-000000000000 does not exist! ValidationError: description: >- The request failed validation. This includes malformed JSON, missing or wrongly typed fields, out-of-range values, and business-rule violations such as text length limits or a voice that is not ready for synthesis. content: application/json: schema: $ref: '#/components/schemas/Error' examples: text_too_long: summary: Text over the 500-character limit value: statusCode: 422 error: ValidationException message: >- Input text of length 812 is more than maximum character count of 500 characters! voice_not_ready: summary: Voice still processing value: statusCode: 422 error: ValidationException message: >- Voice 3acc7441-4dce-4245-a365-0bae822c8c30 is not ready for TTS yet! ServerError: description: >- An unexpected error occurred while processing the request. Any credits charged for a failed generation are refunded. content: application/json: schema: $ref: '#/components/schemas/Error' example: statusCode: 500 error: ProcessException message: TTS prediction error! for input of length 128 securitySchemes: APIKeyHeader: type: apiKey in: header name: chariotai-api-key description: >- Your Chariot AI API key, generated from the dashboard. Send it in the `chariotai-api-key` header on every request. ````