> ## Documentation Index > Fetch the complete documentation index at: https://docs.skills.video/llms.txt > Use this file to discover all available pages before exploring further. # Qwen 3 TTS 1.7B Text-to-speech with preset voices or cloned speaker embeddings and controllable decoding parameters. ## OpenAPI ````yaml /openapi.json post /generation/fal-ai/qwen-3-tts/text-to-speech/1.7b openapi: 3.1.0 info: title: skills.video OpenAPI version: 1.0.0 description: OpenAPI schema for skills.video API endpoints. servers: - url: https://open.skills.video/api/v1 security: [] paths: /generation/fal-ai/qwen-3-tts/text-to-speech/1.7b: post: summary: Qwen 3 TTS 1.7B requestBody: required: true content: application/json: schema: $ref: >- #/components/schemas/qwen_qwen_3_tts_text_to_speech_1_7bCreateRequest multipart/form-data: schema: $ref: >- #/components/schemas/qwen_qwen_3_tts_text_to_speech_1_7bMultipartCreateRequest responses: '200': description: Generation submission accepted content: application/json: example: id: gen_123 status: IN_QUEUE input: prompt: lofi piano with soft rain ambience duration: 15 usage: total: 20 subscription: 20 permanent: 0 schema: $ref: >- #/components/schemas/qwen_qwen_3_tts_text_to_speech_1_7bPrediction '400': description: >- TTS generation request was rejected as invalid or model is not supported. content: application/json: example: error: TTS_GENERATION_INVALID_INPUT code: TTS_GENERATION_INVALID_INPUT message: '''prompt'' parameter is required' schema: anyOf: - $ref: '#/components/schemas/TtsGenerationErrorResponse' - $ref: '#/components/schemas/TemplateErrorResponse' - $ref: '#/components/schemas/ErrorResponse' '401': description: Authentication required or failed content: application/json: example: error: UNAUTHORIZED code: UNAUTHORIZED message: Unauthorized schema: $ref: '#/components/schemas/ErrorResponse' '402': description: Workspace does not have enough credits to run this generation. content: application/json: example: error: INSUFFICIENT_CREDITS code: TTS_GENERATION_INSUFFICIENT_CREDITS message: Insufficient credits for this generation. schema: $ref: '#/components/schemas/TtsGenerationErrorResponse' '403': description: Workspace access is forbidden or Seedance verification is required. content: application/json: example: error: WORKSPACE_FORBIDDEN code: WORKSPACE_FORBIDDEN message: You do not have access to this workspace. schema: anyOf: - $ref: '#/components/schemas/ErrorResponse' - $ref: '#/components/schemas/SeedanceVerificationErrorResponse' '404': description: Resource, model, or template not found. content: application/json: example: error: TEMPLATE_NOT_FOUND code: TEMPLATE_NOT_FOUND message: Template not found. schema: anyOf: - $ref: '#/components/schemas/ErrorResponse' - $ref: '#/components/schemas/TemplateErrorResponse' '422': description: Upstream moderation rejected the request content. content: application/json: example: error: TTS_GENERATION_CONTENT_REJECTED code: TTS_GENERATION_CONTENT_REJECTED message: Request content was rejected by the moderation policy. schema: $ref: '#/components/schemas/TtsGenerationErrorResponse' '429': description: Generation rate limit exceeded upstream. content: application/json: example: error: TTS_GENERATION_RATE_LIMITED code: TTS_GENERATION_RATE_LIMITED message: Rate limit exceeded. Please retry later. schema: $ref: '#/components/schemas/TtsGenerationErrorResponse' '500': description: >- TTS generation failed before provider execution completed. Response includes code and message fields. content: application/json: example: error: TTS_GENERATION_FAILED code: TTS_GENERATION_FAILED message: Generation failed. schema: anyOf: - $ref: '#/components/schemas/TtsGenerationErrorResponse' - $ref: '#/components/schemas/ErrorResponse' '502': description: >- TTS generation failed. Returned when the upstream provider reports a generic failure. content: application/json: example: error: TTS_GENERATION_FAILED code: TTS_GENERATION_FAILED message: Generation failed. schema: $ref: '#/components/schemas/TtsGenerationErrorResponse' '503': description: Upstream generation provider is unavailable. content: application/json: example: error: TTS_GENERATION_PROVIDER_UNAVAILABLE code: TTS_GENERATION_PROVIDER_UNAVAILABLE message: Upstream generation provider is temporarily unavailable. schema: $ref: '#/components/schemas/TtsGenerationErrorResponse' '504': description: TTS generation exceeded the upstream timeout. content: application/json: example: error: TTS_GENERATION_TIMEOUT code: TTS_GENERATION_TIMEOUT message: Generation timed out. schema: $ref: '#/components/schemas/TtsGenerationErrorResponse' security: - bearerAuth: [] components: schemas: qwen_qwen_3_tts_text_to_speech_1_7bCreateRequest: $ref: '#/components/schemas/qwen_qwen_3_tts_text_to_speech_1_7bInput' title: Qwen 3 TTS 1.7B Input qwen_qwen_3_tts_text_to_speech_1_7bMultipartCreateRequest: title: Qwen 3 TTS 1.7B Input type: object properties: text: description: The text to be converted to speech. title: Text examples: - >- I am solving the equation x = (-b +/- sqrt(b^2-4ac)) / 2a. Nobody can, it is a disaster and very sad. order: 1 xPlaceholder: Enter text to convert to speech... xRows: 4 xUiKind: textarea xTemplateRequired: true type: string prompt: description: >- Optional prompt to guide the style of generated speech. Ignored when speaker embedding is provided. title: Style Prompt examples: - Very happy. order: 2 xPlaceholder: e.g. Very happy. xRows: 2 xUiKind: textarea type: string voice: description: >- Preset voice for synthesis. Ignored when speaker embedding is provided. title: Voice examples: - Vivian order: 3 xMeta: voiceProfiles: - id: Vivian language: Chinese languageCode: zh gender: female styleTags: - bright - slightly edgy - young previewUrl: >- https://s.skills.video/6/8_QRF_89_Ijgf_ZL_Uf4_PM_2rf_a12_J34_UA_865d1c918d_773576f4cf.mp3 - id: Serena language: Chinese languageCode: zh gender: female styleTags: - warm - gentle - young previewUrl: >- https://s.skills.video/6/Hlcm_T_Sb_Q_k2_R0d2ow_WS_Lr_MK_Nb5k8_G_315b5a4879_06d7a74a81.mp3 - id: Uncle_Fu language: Chinese languageCode: zh gender: male styleTags: - seasoned - low - mellow previewUrl: >- https://s.skills.video/6/Zq_Xqv_BU_89l7_M3_XKKHN_6_UU_Pwi3m0_a2b27c13cb_50d9bb47a3.mp3 - id: Dylan language: Chinese (Beijing Dialect) languageCode: zh gender: male styleTags: - beijing dialect - youthful - clear previewUrl: >- https://s.skills.video/6/FKO_Zd_Jtg_IR_Ix_Sw0_LM_3_Pc_J_L_Jj_Ljem5_bfd941bdf4_fb07964a16.mp3 - id: Eric language: Chinese (Sichuan Dialect) languageCode: zh gender: male styleTags: - sichuan dialect - lively - husky previewUrl: >- https://s.skills.video/6/x90_Lg8t6kz_PZ_5n_Rxnu_If_F_j9_UQKP_87_02a09f36fd_e82722b118.mp3 - id: Ryan language: English languageCode: en gender: male styleTags: - dynamic - rhythmic - energetic previewUrl: >- https://s.skills.video/6/g_Hw_IY_1_Q8l_B_Mx_CUYD_6_Qd_My_xya_QCPRI_afa60a3b52_4b05c8415e.mp3 - id: Aiden language: English languageCode: en gender: male styleTags: - sunny - american - clear midrange previewUrl: >- https://s.skills.video/6/f_Og_Q4k5hky2_DK_Zr_KCNTA_a_H9_EIM_8i_0d46328589_abdfd46131.mp3 - id: Ono_Anna language: Japanese languageCode: ja gender: female styleTags: - playful - light - nimble previewUrl: >- https://s.skills.video/6/gbryl_Ve_Co7z_GXUI_Mt_Rd_Q_w_LA_Bq_Yq8_9ac8d73014_2ee8418e3a.mp3 - id: Sohee language: Korean languageCode: ko gender: female styleTags: - warm - emotional - rich previewUrl: >- https://s.skills.video/6/W_Ppg_K_Bul_Uk_I_Wyh_Mw_T_Nh_h_E6_P3_O_Cn_3273716cb7_9d5896f0c3.mp3 type: string enum: - Vivian - Serena - Uncle_Fu - Dylan - Eric - Ryan - Aiden - Ono_Anna - Sohee language: description: The language of the voice. title: Language examples: - English order: 4 default: Auto type: string enum: - Auto - English - Chinese - Spanish - French - German - Italian - Japanese - Korean - Portuguese - Russian speaker_voice_embedding_file_url: description: >- URL to a speaker embedding safetensors file from `fal-ai/qwen-3-tts/clone-voice`. If provided, cloned voice is used instead of predefined voices. title: Speaker Embedding URL examples: - >- https://storage.googleapis.com/falserverless/example_outputs/qwen3-tts/clone_out.safetensors format: uri order: 5 xPlaceholder: https://... type: string reference_text: description: >- Optional reference text used when creating the speaker embedding. This can improve quality when using cloned voice. title: Reference Text examples: - >- Okay. Yeah. I resent you. I love you. I respect you. But you know what? You blew it! order: 6 xRows: 3 xUiKind: textarea type: string top_k: description: Top-k sampling parameter. title: Top K order: 7 xStep: 1 default: 50 type: integer minimum: 0 maximum: 9007199254740991 top_p: description: Top-p sampling parameter. title: Top P order: 8 xStep: 0.01 default: 1 type: number minimum: 0 maximum: 1 temperature: description: Sampling temperature; higher means more random output. title: Temperature order: 9 xStep: 0.01 default: 0.9 type: number minimum: 0 maximum: 1 repetition_penalty: description: Penalty to reduce repeated tokens/codes. title: Repetition Penalty order: 10 xStep: 0.01 default: 1.05 type: number minimum: 0 subtalker_dosample: description: Sampling switch for the sub-talker. title: Subtalker Do Sample order: 11 default: true type: boolean subtalker_top_k: description: Top-k sampling for the sub-talker. title: Subtalker Top K order: 12 xStep: 1 default: 50 type: integer minimum: 0 maximum: 9007199254740991 subtalker_top_p: description: Top-p for sub-talker sampling. title: Subtalker Top P order: 13 xStep: 0.01 default: 1 type: number minimum: 0 maximum: 1 subtalker_temperature: description: Temperature for sub-talker sampling. title: Subtalker Temperature order: 14 xStep: 0.01 default: 0.9 type: number minimum: 0 maximum: 1 max_new_tokens: description: Maximum number of new codec tokens to generate. title: Max New Tokens order: 15 xStep: 1 default: 200 type: integer minimum: 1 maximum: 8192 required: - text additionalProperties: false qwen_qwen_3_tts_text_to_speech_1_7bPrediction: type: object properties: id: description: Generation id. type: string status: $ref: '#/components/schemas/QueueState' input: $ref: '#/components/schemas/qwen_qwen_3_tts_text_to_speech_1_7bInputOutput' title: Qwen 3 TTS 1.7B Input description: Normalized input payload for the generation task. usage: $ref: '#/components/schemas/Usage' code: $ref: '#/components/schemas/__schema1' message: $ref: '#/components/schemas/__schema2' required: - id - status - input - usage additionalProperties: false TtsGenerationErrorResponse: type: object properties: error: description: >- Legacy machine-readable error field. Prefer `code` for new integrations. type: string code: description: >- TTS generation error code. `*_FAILED` is the generic fallback; more specific codes (`*_TIMEOUT`, `*_RATE_LIMITED`, `*_PROVIDER_UNAVAILABLE`, `*_UNSUPPORTED_MODEL`, `*_INVALID_INPUT`, `*_INSUFFICIENT_CREDITS`, `*_CONTENT_REJECTED`) are surfaced when applicable. type: string enum: - TTS_GENERATION_FAILED - TTS_GENERATION_TIMEOUT - TTS_GENERATION_RATE_LIMITED - TTS_GENERATION_PROVIDER_UNAVAILABLE - TTS_GENERATION_UNSUPPORTED_MODEL - TTS_GENERATION_INVALID_INPUT - TTS_GENERATION_INSUFFICIENT_CREDITS - TTS_GENERATION_CONTENT_REJECTED message: description: Human readable error message. Upstream vendor names are redacted. type: string errors: description: Field-level validation errors when available. type: array items: type: object properties: path: type: string message: type: string code: type: string required: - path - message - code additionalProperties: false details: description: Additional structured error details when available. requested: description: Requested credits when the error is insufficient credits. type: number available: description: Available credits when the error is insufficient credits. type: number required: - code - message additionalProperties: false TemplateErrorResponse: type: object properties: error: description: >- Legacy machine-readable error field. Prefer `code` for new integrations. type: string code: description: Template error code. type: string enum: - TEMPLATE_NOT_FOUND - TEMPLATE_MODEL_MISMATCH - TEMPLATE_INVALID_INPUT - TEMPLATE_TRANSFORM_FAILED message: description: Human readable template error message. type: string errors: description: Field-level template validation errors when available. type: array items: type: object properties: path: type: string message: type: string code: type: string required: - path - message - code additionalProperties: false details: description: Additional structured error details when available. required: - code - message additionalProperties: false ErrorResponse: type: object properties: error: description: >- Legacy machine-readable error field. Prefer `code` for new integrations. This may contain an unprefixed compatibility code for older clients. type: string code: description: Machine-readable error code. Use this to branch in client code. type: string message: description: Human readable error message. type: string errors: description: Field-level validation errors when available. type: array items: type: object properties: path: type: string message: type: string code: type: string required: - path - message - code additionalProperties: false details: description: Additional structured error details when available. required: - code - message additionalProperties: false SeedanceVerificationErrorResponse: type: object properties: error: description: >- Legacy machine-readable error field. Prefer `code` for new integrations. type: string code: description: Seedance verification error code. type: string const: SEEDANCE_VERIFICATION_REQUIRED message: description: Human readable verification error message. type: string verification: description: Current Seedance verification state when available. anyOf: - {} - type: 'null' required: - code - message additionalProperties: false qwen_qwen_3_tts_text_to_speech_1_7bInput: title: Qwen 3 TTS 1.7B Input type: object properties: text: description: The text to be converted to speech. title: Text examples: - >- I am solving the equation x = (-b +/- sqrt(b^2-4ac)) / 2a. Nobody can, it is a disaster and very sad. order: 1 xPlaceholder: Enter text to convert to speech... xRows: 4 xUiKind: textarea xTemplateRequired: true type: string prompt: description: >- Optional prompt to guide the style of generated speech. Ignored when speaker embedding is provided. title: Style Prompt examples: - Very happy. order: 2 xPlaceholder: e.g. Very happy. xRows: 2 xUiKind: textarea type: string voice: description: >- Preset voice for synthesis. Ignored when speaker embedding is provided. title: Voice examples: - Vivian order: 3 xMeta: voiceProfiles: - id: Vivian language: Chinese languageCode: zh gender: female styleTags: - bright - slightly edgy - young previewUrl: >- https://s.skills.video/6/8_QRF_89_Ijgf_ZL_Uf4_PM_2rf_a12_J34_UA_865d1c918d_773576f4cf.mp3 - id: Serena language: Chinese languageCode: zh gender: female styleTags: - warm - gentle - young previewUrl: >- https://s.skills.video/6/Hlcm_T_Sb_Q_k2_R0d2ow_WS_Lr_MK_Nb5k8_G_315b5a4879_06d7a74a81.mp3 - id: Uncle_Fu language: Chinese languageCode: zh gender: male styleTags: - seasoned - low - mellow previewUrl: >- https://s.skills.video/6/Zq_Xqv_BU_89l7_M3_XKKHN_6_UU_Pwi3m0_a2b27c13cb_50d9bb47a3.mp3 - id: Dylan language: Chinese (Beijing Dialect) languageCode: zh gender: male styleTags: - beijing dialect - youthful - clear previewUrl: >- https://s.skills.video/6/FKO_Zd_Jtg_IR_Ix_Sw0_LM_3_Pc_J_L_Jj_Ljem5_bfd941bdf4_fb07964a16.mp3 - id: Eric language: Chinese (Sichuan Dialect) languageCode: zh gender: male styleTags: - sichuan dialect - lively - husky previewUrl: >- https://s.skills.video/6/x90_Lg8t6kz_PZ_5n_Rxnu_If_F_j9_UQKP_87_02a09f36fd_e82722b118.mp3 - id: Ryan language: English languageCode: en gender: male styleTags: - dynamic - rhythmic - energetic previewUrl: >- https://s.skills.video/6/g_Hw_IY_1_Q8l_B_Mx_CUYD_6_Qd_My_xya_QCPRI_afa60a3b52_4b05c8415e.mp3 - id: Aiden language: English languageCode: en gender: male styleTags: - sunny - american - clear midrange previewUrl: >- https://s.skills.video/6/f_Og_Q4k5hky2_DK_Zr_KCNTA_a_H9_EIM_8i_0d46328589_abdfd46131.mp3 - id: Ono_Anna language: Japanese languageCode: ja gender: female styleTags: - playful - light - nimble previewUrl: >- https://s.skills.video/6/gbryl_Ve_Co7z_GXUI_Mt_Rd_Q_w_LA_Bq_Yq8_9ac8d73014_2ee8418e3a.mp3 - id: Sohee language: Korean languageCode: ko gender: female styleTags: - warm - emotional - rich previewUrl: >- https://s.skills.video/6/W_Ppg_K_Bul_Uk_I_Wyh_Mw_T_Nh_h_E6_P3_O_Cn_3273716cb7_9d5896f0c3.mp3 type: string enum: - Vivian - Serena - Uncle_Fu - Dylan - Eric - Ryan - Aiden - Ono_Anna - Sohee language: description: The language of the voice. title: Language examples: - English order: 4 default: Auto type: string enum: - Auto - English - Chinese - Spanish - French - German - Italian - Japanese - Korean - Portuguese - Russian speaker_voice_embedding_file_url: description: >- URL to a speaker embedding safetensors file from `fal-ai/qwen-3-tts/clone-voice`. If provided, cloned voice is used instead of predefined voices. title: Speaker Embedding URL examples: - >- https://storage.googleapis.com/falserverless/example_outputs/qwen3-tts/clone_out.safetensors format: uri order: 5 xPlaceholder: https://... type: string reference_text: description: >- Optional reference text used when creating the speaker embedding. This can improve quality when using cloned voice. title: Reference Text examples: - >- Okay. Yeah. I resent you. I love you. I respect you. But you know what? You blew it! order: 6 xRows: 3 xUiKind: textarea type: string top_k: description: Top-k sampling parameter. title: Top K order: 7 xStep: 1 default: 50 type: integer minimum: 0 maximum: 9007199254740991 top_p: description: Top-p sampling parameter. title: Top P order: 8 xStep: 0.01 default: 1 type: number minimum: 0 maximum: 1 temperature: description: Sampling temperature; higher means more random output. title: Temperature order: 9 xStep: 0.01 default: 0.9 type: number minimum: 0 maximum: 1 repetition_penalty: description: Penalty to reduce repeated tokens/codes. title: Repetition Penalty order: 10 xStep: 0.01 default: 1.05 type: number minimum: 0 subtalker_dosample: description: Sampling switch for the sub-talker. title: Subtalker Do Sample order: 11 default: true type: boolean subtalker_top_k: description: Top-k sampling for the sub-talker. title: Subtalker Top K order: 12 xStep: 1 default: 50 type: integer minimum: 0 maximum: 9007199254740991 subtalker_top_p: description: Top-p for sub-talker sampling. title: Subtalker Top P order: 13 xStep: 0.01 default: 1 type: number minimum: 0 maximum: 1 subtalker_temperature: description: Temperature for sub-talker sampling. title: Subtalker Temperature order: 14 xStep: 0.01 default: 0.9 type: number minimum: 0 maximum: 1 max_new_tokens: description: Maximum number of new codec tokens to generate. title: Max New Tokens order: 15 xStep: 1 default: 200 type: integer minimum: 1 maximum: 8192 required: - text additionalProperties: false QueueState: description: Queue status type: string enum: - IN_QUEUE - IN_PROGRESS - COMPLETED - FAILED - CANCELED qwen_qwen_3_tts_text_to_speech_1_7bInputOutput: $ref: '#/components/schemas/__schema70' title: Qwen 3 TTS 1.7B Input Usage: description: Credit usage breakdown for the request type: object properties: total: type: integer minimum: -9007199254740991 maximum: 9007199254740991 subscription: type: integer minimum: -9007199254740991 maximum: 9007199254740991 permanent: type: integer minimum: -9007199254740991 maximum: 9007199254740991 required: - total - subscription - permanent additionalProperties: false __schema1: description: >- Machine-readable generation error code. Present when `status` is `FAILED`. type: string __schema2: description: >- Sanitized human-readable failure message. Present when `status` is `FAILED`. type: string __schema70: title: Qwen 3 TTS 1.7B Input type: object properties: text: description: The text to be converted to speech. title: Text examples: - >- I am solving the equation x = (-b +/- sqrt(b^2-4ac)) / 2a. Nobody can, it is a disaster and very sad. order: 1 xPlaceholder: Enter text to convert to speech... xRows: 4 xUiKind: textarea xTemplateRequired: true type: string prompt: description: >- Optional prompt to guide the style of generated speech. Ignored when speaker embedding is provided. title: Style Prompt examples: - Very happy. order: 2 xPlaceholder: e.g. Very happy. xRows: 2 xUiKind: textarea type: string voice: description: >- Preset voice for synthesis. Ignored when speaker embedding is provided. title: Voice examples: - Vivian order: 3 xMeta: voiceProfiles: - id: Vivian language: Chinese languageCode: zh gender: female styleTags: - bright - slightly edgy - young previewUrl: >- https://s.skills.video/6/8_QRF_89_Ijgf_ZL_Uf4_PM_2rf_a12_J34_UA_865d1c918d_773576f4cf.mp3 - id: Serena language: Chinese languageCode: zh gender: female styleTags: - warm - gentle - young previewUrl: >- https://s.skills.video/6/Hlcm_T_Sb_Q_k2_R0d2ow_WS_Lr_MK_Nb5k8_G_315b5a4879_06d7a74a81.mp3 - id: Uncle_Fu language: Chinese languageCode: zh gender: male styleTags: - seasoned - low - mellow previewUrl: >- https://s.skills.video/6/Zq_Xqv_BU_89l7_M3_XKKHN_6_UU_Pwi3m0_a2b27c13cb_50d9bb47a3.mp3 - id: Dylan language: Chinese (Beijing Dialect) languageCode: zh gender: male styleTags: - beijing dialect - youthful - clear previewUrl: >- https://s.skills.video/6/FKO_Zd_Jtg_IR_Ix_Sw0_LM_3_Pc_J_L_Jj_Ljem5_bfd941bdf4_fb07964a16.mp3 - id: Eric language: Chinese (Sichuan Dialect) languageCode: zh gender: male styleTags: - sichuan dialect - lively - husky previewUrl: >- https://s.skills.video/6/x90_Lg8t6kz_PZ_5n_Rxnu_If_F_j9_UQKP_87_02a09f36fd_e82722b118.mp3 - id: Ryan language: English languageCode: en gender: male styleTags: - dynamic - rhythmic - energetic previewUrl: >- https://s.skills.video/6/g_Hw_IY_1_Q8l_B_Mx_CUYD_6_Qd_My_xya_QCPRI_afa60a3b52_4b05c8415e.mp3 - id: Aiden language: English languageCode: en gender: male styleTags: - sunny - american - clear midrange previewUrl: >- https://s.skills.video/6/f_Og_Q4k5hky2_DK_Zr_KCNTA_a_H9_EIM_8i_0d46328589_abdfd46131.mp3 - id: Ono_Anna language: Japanese languageCode: ja gender: female styleTags: - playful - light - nimble previewUrl: >- https://s.skills.video/6/gbryl_Ve_Co7z_GXUI_Mt_Rd_Q_w_LA_Bq_Yq8_9ac8d73014_2ee8418e3a.mp3 - id: Sohee language: Korean languageCode: ko gender: female styleTags: - warm - emotional - rich previewUrl: >- https://s.skills.video/6/W_Ppg_K_Bul_Uk_I_Wyh_Mw_T_Nh_h_E6_P3_O_Cn_3273716cb7_9d5896f0c3.mp3 type: string enum: - Vivian - Serena - Uncle_Fu - Dylan - Eric - Ryan - Aiden - Ono_Anna - Sohee language: description: The language of the voice. title: Language examples: - English order: 4 default: Auto type: string enum: - Auto - English - Chinese - Spanish - French - German - Italian - Japanese - Korean - Portuguese - Russian speaker_voice_embedding_file_url: description: >- URL to a speaker embedding safetensors file from `fal-ai/qwen-3-tts/clone-voice`. If provided, cloned voice is used instead of predefined voices. title: Speaker Embedding URL examples: - >- https://storage.googleapis.com/falserverless/example_outputs/qwen3-tts/clone_out.safetensors format: uri order: 5 xPlaceholder: https://... type: string reference_text: description: >- Optional reference text used when creating the speaker embedding. This can improve quality when using cloned voice. title: Reference Text examples: - >- Okay. Yeah. I resent you. I love you. I respect you. But you know what? You blew it! order: 6 xRows: 3 xUiKind: textarea type: string top_k: description: Top-k sampling parameter. title: Top K order: 7 xStep: 1 default: 50 type: integer minimum: 0 maximum: 9007199254740991 top_p: description: Top-p sampling parameter. title: Top P order: 8 xStep: 0.01 default: 1 type: number minimum: 0 maximum: 1 temperature: description: Sampling temperature; higher means more random output. title: Temperature order: 9 xStep: 0.01 default: 0.9 type: number minimum: 0 maximum: 1 repetition_penalty: description: Penalty to reduce repeated tokens/codes. title: Repetition Penalty order: 10 xStep: 0.01 default: 1.05 type: number minimum: 0 subtalker_dosample: description: Sampling switch for the sub-talker. title: Subtalker Do Sample order: 11 default: true type: boolean subtalker_top_k: description: Top-k sampling for the sub-talker. title: Subtalker Top K order: 12 xStep: 1 default: 50 type: integer minimum: 0 maximum: 9007199254740991 subtalker_top_p: description: Top-p for sub-talker sampling. title: Subtalker Top P order: 13 xStep: 0.01 default: 1 type: number minimum: 0 maximum: 1 subtalker_temperature: description: Temperature for sub-talker sampling. title: Subtalker Temperature order: 14 xStep: 0.01 default: 0.9 type: number minimum: 0 maximum: 1 max_new_tokens: description: Maximum number of new codec tokens to generate. title: Max New Tokens order: 15 xStep: 1 default: 200 type: integer minimum: 1 maximum: 8192 required: - text additionalProperties: false securitySchemes: bearerAuth: type: http scheme: bearer bearerFormat: API Key description: API Key for API endpoints ````