ai 4.3.5 → 4.3.6

package/CHANGELOG.md

@@ -1,5 +1,17 @@
 # ai
 
+## 4.3.6
+
+### Patch Changes
+
+- beef951: feat: add speech with experimental_generateSpeech
+- bd41167: fix(ai/core): properly handle custom separator in provider registry
+- Updated dependencies [beef951]
+  - @ai-sdk/provider@1.1.3
+  - @ai-sdk/provider-utils@2.2.7
+  - @ai-sdk/ui-utils@1.2.8
+  - @ai-sdk/react@1.2.9
+
 ## 4.3.5
 
 ### Patch Changes

package/dist/index.d.mts

@@ -2,7 +2,7 @@ import { IDGenerator } from '@ai-sdk/provider-utils';
 export { CoreToolCall, CoreToolResult, IDGenerator, ToolCall, ToolResult, createIdGenerator, generateId } from '@ai-sdk/provider-utils';
 import { DataStreamString, Message, Schema, DeepPartial, JSONValue as JSONValue$1, AssistantMessage, DataMessage } from '@ai-sdk/ui-utils';
 export { AssistantMessage, AssistantStatus, Attachment, ChatRequest, ChatRequestOptions, CreateMessage, DataMessage, DataStreamPart, DeepPartial, IdGenerator, JSONValue, Message, RequestOptions, Schema, ToolInvocation, UIMessage, UseAssistantOptions, formatAssistantStreamPart, formatDataStreamPart, jsonSchema, parseAssistantStreamPart, parseDataStreamPart, processDataStream, processTextStream, zodSchema } from '@ai-sdk/ui-utils';
-import { LanguageModelV1, LanguageModelV1FinishReason, LanguageModelV1LogProbs, LanguageModelV1CallWarning, LanguageModelV1Source, JSONValue, EmbeddingModelV1, EmbeddingModelV1Embedding, ImageModelV1, ImageModelV1CallWarning, LanguageModelV1ProviderMetadata, LanguageModelV1CallOptions, AISDKError, LanguageModelV1FunctionToolCall, JSONSchema7, JSONParseError, TypeValidationError, TranscriptionModelV1CallWarning, TranscriptionModelV1, ProviderV1, NoSuchModelError } from '@ai-sdk/provider';
+import { LanguageModelV1, LanguageModelV1FinishReason, LanguageModelV1LogProbs, LanguageModelV1CallWarning, LanguageModelV1Source, JSONValue, EmbeddingModelV1, EmbeddingModelV1Embedding, ImageModelV1, ImageModelV1CallWarning, LanguageModelV1ProviderMetadata, TranscriptionModelV1, TranscriptionModelV1CallWarning, SpeechModelV1, SpeechModelV1CallWarning, LanguageModelV1CallOptions, AISDKError, LanguageModelV1FunctionToolCall, JSONSchema7, JSONParseError, TypeValidationError, ProviderV1, NoSuchModelError } from '@ai-sdk/provider';
 export { AISDKError, APICallError, EmptyResponseBodyError, InvalidPromptError, InvalidResponseDataError, JSONParseError, LanguageModelV1, LanguageModelV1CallOptions, LanguageModelV1Prompt, LanguageModelV1StreamPart, LoadAPIKeyError, NoContentGeneratedError, NoSuchModelError, TypeValidationError, UnsupportedFunctionalityError } from '@ai-sdk/provider';
 import { ServerResponse } from 'node:http';
 import { AttributeValue, Tracer } from '@opentelemetry/api';
@@ -276,6 +276,56 @@ type EmbeddingModelUsage = {
     tokens: number;
 };
 
+/**
+Transcription model that is used by the AI SDK Core functions.
+  */
+type TranscriptionModel = TranscriptionModelV1;
+/**
+Warning from the model provider for this call. The call will proceed, but e.g.
+some settings might not be supported, which can lead to suboptimal results.
+  */
+type TranscriptionWarning = TranscriptionModelV1CallWarning;
+
+type TranscriptionModelResponseMetadata = {
+    /**
+  Timestamp for the start of the generated response.
+     */
+    timestamp: Date;
+    /**
+  The ID of the response model that was used to generate the response.
+     */
+    modelId: string;
+    /**
+  Response headers.
+     */
+    headers?: Record<string, string>;
+};
+
+/**
+Speech model that is used by the AI SDK Core functions.
+  */
+type SpeechModel = SpeechModelV1;
+/**
+Warning from the model provider for this call. The call will proceed, but e.g.
+some settings might not be supported, which can lead to suboptimal results.
+  */
+type SpeechWarning = SpeechModelV1CallWarning;
+
+type SpeechModelResponseMetadata = {
+    /**
+  Timestamp for the start of the generated response.
+     */
+    timestamp: Date;
+    /**
+  The ID of the response model that was used to generate the response.
+     */
+    modelId: string;
+    /**
+  Response headers.
+     */
+    headers?: Record<string, string>;
+};
+
 /**
 The result of an `embed` call.
 It contains the embedding, the value, and additional information.
@@ -3793,25 +3843,109 @@ Callback that is called when the LLM response and the final object validation ar
 }): StreamObjectResult<JSONValue, JSONValue, never>;
 
 /**
-Warning from the model provider for this call. The call will proceed, but e.g.
-some settings might not be supported, which can lead to suboptimal results.
-  */
-type TranscriptionWarning = TranscriptionModelV1CallWarning;
+ * A generated audio file.
+ */
+interface GeneratedAudioFile extends GeneratedFile {
+    /**
+     * Audio format of the file (e.g., 'mp3', 'wav', etc.)
+     */
+    readonly format: string;
+}
 
-type TranscriptionModelResponseMetadata = {
+/**
+The result of a `generateSpeech` call.
+It contains the audio data and additional information.
+ */
+interface SpeechResult {
     /**
-  Timestamp for the start of the generated response.
+     * The audio data as a base64 encoded string or binary data.
      */
-    timestamp: Date;
+    readonly audio: GeneratedAudioFile;
     /**
-  The ID of the response model that was used to generate the response.
+    Warnings for the call, e.g. unsupported settings.
+       */
+    readonly warnings: Array<SpeechWarning>;
+    /**
+    Response metadata from the provider. There may be multiple responses if we made multiple calls to the model.
      */
-    modelId: string;
+    readonly responses: Array<SpeechModelResponseMetadata>;
     /**
-  Response headers.
+    Provider metadata from the provider.
      */
+    readonly providerMetadata: Record<string, Record<string, JSONValue>>;
+}
+
+/**
+Generates speech audio using a speech model.
+
+@param model - The speech model to use.
+@param text - The text to convert to speech.
+@param voice - The voice to use for speech generation.
+@param outputFormat - The output format to use for speech generation e.g. "mp3", "wav", etc.
+@param instructions - Instructions for the speech generation e.g. "Speak in a slow and steady tone".
+@param speed - The speed of the speech generation.
+@param providerOptions - Additional provider-specific options that are passed through to the provider
+as body parameters.
+@param maxRetries - Maximum number of retries. Set to 0 to disable retries. Default: 2.
+@param abortSignal - An optional abort signal that can be used to cancel the call.
+@param headers - Additional HTTP headers to be sent with the request. Only applicable for HTTP-based providers.
+
+@returns A result object that contains the generated audio data.
+ */
+declare function generateSpeech({ model, text, voice, outputFormat, instructions, speed, providerOptions, maxRetries: maxRetriesArg, abortSignal, headers, }: {
+    /**
+  The speech model to use.
+       */
+    model: SpeechModelV1;
+    /**
+  The text to convert to speech.
+     */
+    text: string;
+    /**
+  The voice to use for speech generation.
+     */
+    voice?: string;
+    /**
+     * The desired output format for the audio e.g. "mp3", "wav", etc.
+     */
+    outputFormat?: 'mp3' | 'wav' | (string & {});
+    /**
+      Instructions for the speech generation e.g. "Speak in a slow and steady tone".
+    */
+    instructions?: string;
+    /**
+    The speed of the speech generation.
+     */
+    speed?: number;
+    /**
+  Additional provider-specific options that are passed through to the provider
+  as body parameters.
+  
+  The outer record is keyed by the provider name, and the inner
+  record is keyed by the provider-specific metadata key.
+  ```ts
+  {
+    "openai": {}
+  }
+  ```
+       */
+    providerOptions?: ProviderOptions;
+    /**
+  Maximum number of retries per speech model call. Set to 0 to disable retries.
+  
+  @default 2
+     */
+    maxRetries?: number;
+    /**
+  Abort signal.
+   */
+    abortSignal?: AbortSignal;
+    /**
+  Additional headers to include in the request.
+  Only applicable for HTTP-based providers.
+   */
     headers?: Record<string, string>;
-};
+}): Promise<SpeechResult>;
 
 /**
 The result of a `transcribe` call.
@@ -4521,4 +4655,4 @@ declare namespace llamaindexAdapter {
   };
 }
 
-export { AssistantContent, AssistantResponse, CallWarning, ChunkDetector, CoreAssistantMessage, CoreMessage, CoreSystemMessage, CoreTool, CoreToolCallUnion, CoreToolChoice, CoreToolMessage, CoreToolResultUnion, CoreUserMessage, DataContent, DataStreamOptions, DataStreamWriter, DownloadError, EmbedManyResult, EmbedResult, Embedding, EmbeddingModel, EmbeddingModelUsage, GenerateImageResult as Experimental_GenerateImageResult, GeneratedFile as Experimental_GeneratedImage, Experimental_LanguageModelV1Middleware, TranscriptionResult as Experimental_TranscriptionResult, FilePart, FinishReason, GenerateObjectResult, GenerateTextOnStepFinishCallback, GenerateTextResult, GeneratedFile, ImageModel, ImageGenerationWarning as ImageModelCallWarning, ImageModelResponseMetadata, ImagePart, InvalidArgumentError, InvalidDataContentError, InvalidMessageRoleError, InvalidStreamPartError, InvalidToolArgumentsError, JSONRPCError, JSONRPCMessage, JSONRPCNotification, JSONRPCRequest, JSONRPCResponse, langchainAdapter as LangChainAdapter, LanguageModel, LanguageModelRequestMetadata, LanguageModelResponseMetadata, LanguageModelUsage, LanguageModelV1Middleware, llamaindexAdapter as LlamaIndexAdapter, LogProbs, MCPClientError, MCPTransport, MessageConversionError, NoImageGeneratedError, NoObjectGeneratedError, NoOutputSpecifiedError, NoSuchProviderError, NoSuchToolError, ObjectStreamPart, output as Output, Provider, ProviderMetadata, ProviderRegistryProvider, RepairTextFunction, RetryError, StepResult, StreamData, StreamObjectOnFinishCallback, StreamObjectResult, StreamTextOnChunkCallback, StreamTextOnErrorCallback, StreamTextOnFinishCallback, StreamTextOnStepFinishCallback, StreamTextResult, StreamTextTransform, TelemetrySettings, TextPart, TextStreamPart, Tool, ToolCallPart, ToolCallRepairError, ToolCallRepairFunction, ToolCallUnion, ToolChoice, ToolContent, ToolExecutionError, ToolExecutionOptions, ToolResultPart, ToolResultUnion, ToolSet, UserContent, appendClientMessage, appendResponseMessages, convertToCoreMessages, coreAssistantMessageSchema, coreMessageSchema, coreSystemMessageSchema, coreToolMessageSchema, coreUserMessageSchema, cosineSimilarity, createDataStream, createDataStreamResponse, createProviderRegistry, customProvider, defaultSettingsMiddleware, embed, embedMany, createMCPClient as experimental_createMCPClient, experimental_createProviderRegistry, experimental_customProvider, generateImage as experimental_generateImage, transcribe as experimental_transcribe, experimental_wrapLanguageModel, extractReasoningMiddleware, generateObject, generateText, pipeDataStreamToResponse, simulateReadableStream, simulateStreamingMiddleware, smoothStream, streamObject, streamText, tool, wrapLanguageModel };
+export { AssistantContent, AssistantResponse, CallWarning, ChunkDetector, CoreAssistantMessage, CoreMessage, CoreSystemMessage, CoreTool, CoreToolCallUnion, CoreToolChoice, CoreToolMessage, CoreToolResultUnion, CoreUserMessage, DataContent, DataStreamOptions, DataStreamWriter, DownloadError, EmbedManyResult, EmbedResult, Embedding, EmbeddingModel, EmbeddingModelUsage, GenerateImageResult as Experimental_GenerateImageResult, GeneratedFile as Experimental_GeneratedImage, Experimental_LanguageModelV1Middleware, SpeechResult as Experimental_SpeechResult, TranscriptionResult as Experimental_TranscriptionResult, FilePart, FinishReason, GenerateObjectResult, GenerateTextOnStepFinishCallback, GenerateTextResult, GeneratedAudioFile, GeneratedFile, ImageModel, ImageGenerationWarning as ImageModelCallWarning, ImageModelResponseMetadata, ImagePart, InvalidArgumentError, InvalidDataContentError, InvalidMessageRoleError, InvalidStreamPartError, InvalidToolArgumentsError, JSONRPCError, JSONRPCMessage, JSONRPCNotification, JSONRPCRequest, JSONRPCResponse, langchainAdapter as LangChainAdapter, LanguageModel, LanguageModelRequestMetadata, LanguageModelResponseMetadata, LanguageModelUsage, LanguageModelV1Middleware, llamaindexAdapter as LlamaIndexAdapter, LogProbs, MCPClientError, MCPTransport, MessageConversionError, NoImageGeneratedError, NoObjectGeneratedError, NoOutputSpecifiedError, NoSuchProviderError, NoSuchToolError, ObjectStreamPart, output as Output, Provider, ProviderMetadata, ProviderRegistryProvider, RepairTextFunction, RetryError, SpeechModel, SpeechModelResponseMetadata, SpeechWarning, StepResult, StreamData, StreamObjectOnFinishCallback, StreamObjectResult, StreamTextOnChunkCallback, StreamTextOnErrorCallback, StreamTextOnFinishCallback, StreamTextOnStepFinishCallback, StreamTextResult, StreamTextTransform, TelemetrySettings, TextPart, TextStreamPart, Tool, ToolCallPart, ToolCallRepairError, ToolCallRepairFunction, ToolCallUnion, ToolChoice, ToolContent, ToolExecutionError, ToolExecutionOptions, ToolResultPart, ToolResultUnion, ToolSet, TranscriptionModel, TranscriptionModelResponseMetadata, TranscriptionWarning, UserContent, appendClientMessage, appendResponseMessages, convertToCoreMessages, coreAssistantMessageSchema, coreMessageSchema, coreSystemMessageSchema, coreToolMessageSchema, coreUserMessageSchema, cosineSimilarity, createDataStream, createDataStreamResponse, createProviderRegistry, customProvider, defaultSettingsMiddleware, embed, embedMany, createMCPClient as experimental_createMCPClient, experimental_createProviderRegistry, experimental_customProvider, generateImage as experimental_generateImage, generateSpeech as experimental_generateSpeech, transcribe as experimental_transcribe, experimental_wrapLanguageModel, extractReasoningMiddleware, generateObject, generateText, pipeDataStreamToResponse, simulateReadableStream, simulateStreamingMiddleware, smoothStream, streamObject, streamText, tool, wrapLanguageModel };

package/dist/index.d.ts

@@ -2,7 +2,7 @@ import { IDGenerator } from '@ai-sdk/provider-utils';
 export { CoreToolCall, CoreToolResult, IDGenerator, ToolCall, ToolResult, createIdGenerator, generateId } from '@ai-sdk/provider-utils';
 import { DataStreamString, Message, Schema, DeepPartial, JSONValue as JSONValue$1, AssistantMessage, DataMessage } from '@ai-sdk/ui-utils';
 export { AssistantMessage, AssistantStatus, Attachment, ChatRequest, ChatRequestOptions, CreateMessage, DataMessage, DataStreamPart, DeepPartial, IdGenerator, JSONValue, Message, RequestOptions, Schema, ToolInvocation, UIMessage, UseAssistantOptions, formatAssistantStreamPart, formatDataStreamPart, jsonSchema, parseAssistantStreamPart, parseDataStreamPart, processDataStream, processTextStream, zodSchema } from '@ai-sdk/ui-utils';
-import { LanguageModelV1, LanguageModelV1FinishReason, LanguageModelV1LogProbs, LanguageModelV1CallWarning, LanguageModelV1Source, JSONValue, EmbeddingModelV1, EmbeddingModelV1Embedding, ImageModelV1, ImageModelV1CallWarning, LanguageModelV1ProviderMetadata, LanguageModelV1CallOptions, AISDKError, LanguageModelV1FunctionToolCall, JSONSchema7, JSONParseError, TypeValidationError, TranscriptionModelV1CallWarning, TranscriptionModelV1, ProviderV1, NoSuchModelError } from '@ai-sdk/provider';
+import { LanguageModelV1, LanguageModelV1FinishReason, LanguageModelV1LogProbs, LanguageModelV1CallWarning, LanguageModelV1Source, JSONValue, EmbeddingModelV1, EmbeddingModelV1Embedding, ImageModelV1, ImageModelV1CallWarning, LanguageModelV1ProviderMetadata, TranscriptionModelV1, TranscriptionModelV1CallWarning, SpeechModelV1, SpeechModelV1CallWarning, LanguageModelV1CallOptions, AISDKError, LanguageModelV1FunctionToolCall, JSONSchema7, JSONParseError, TypeValidationError, ProviderV1, NoSuchModelError } from '@ai-sdk/provider';
 export { AISDKError, APICallError, EmptyResponseBodyError, InvalidPromptError, InvalidResponseDataError, JSONParseError, LanguageModelV1, LanguageModelV1CallOptions, LanguageModelV1Prompt, LanguageModelV1StreamPart, LoadAPIKeyError, NoContentGeneratedError, NoSuchModelError, TypeValidationError, UnsupportedFunctionalityError } from '@ai-sdk/provider';
 import { ServerResponse } from 'node:http';
 import { AttributeValue, Tracer } from '@opentelemetry/api';
@@ -276,6 +276,56 @@ type EmbeddingModelUsage = {
     tokens: number;
 };
 
+/**
+Transcription model that is used by the AI SDK Core functions.
+  */
+type TranscriptionModel = TranscriptionModelV1;
+/**
+Warning from the model provider for this call. The call will proceed, but e.g.
+some settings might not be supported, which can lead to suboptimal results.
+  */
+type TranscriptionWarning = TranscriptionModelV1CallWarning;
+
+type TranscriptionModelResponseMetadata = {
+    /**
+  Timestamp for the start of the generated response.
+     */
+    timestamp: Date;
+    /**
+  The ID of the response model that was used to generate the response.
+     */
+    modelId: string;
+    /**
+  Response headers.
+     */
+    headers?: Record<string, string>;
+};
+
+/**
+Speech model that is used by the AI SDK Core functions.
+  */
+type SpeechModel = SpeechModelV1;
+/**
+Warning from the model provider for this call. The call will proceed, but e.g.
+some settings might not be supported, which can lead to suboptimal results.
+  */
+type SpeechWarning = SpeechModelV1CallWarning;
+
+type SpeechModelResponseMetadata = {
+    /**
+  Timestamp for the start of the generated response.
+     */
+    timestamp: Date;
+    /**
+  The ID of the response model that was used to generate the response.
+     */
+    modelId: string;
+    /**
+  Response headers.
+     */
+    headers?: Record<string, string>;
+};
+
 /**
 The result of an `embed` call.
 It contains the embedding, the value, and additional information.
@@ -3793,25 +3843,109 @@ Callback that is called when the LLM response and the final object validation ar
 }): StreamObjectResult<JSONValue, JSONValue, never>;
 
 /**
-Warning from the model provider for this call. The call will proceed, but e.g.
-some settings might not be supported, which can lead to suboptimal results.
-  */
-type TranscriptionWarning = TranscriptionModelV1CallWarning;
+ * A generated audio file.
+ */
+interface GeneratedAudioFile extends GeneratedFile {
+    /**
+     * Audio format of the file (e.g., 'mp3', 'wav', etc.)
+     */
+    readonly format: string;
+}
 
-type TranscriptionModelResponseMetadata = {
+/**
+The result of a `generateSpeech` call.
+It contains the audio data and additional information.
+ */
+interface SpeechResult {
     /**
-  Timestamp for the start of the generated response.
+     * The audio data as a base64 encoded string or binary data.
      */
-    timestamp: Date;
+    readonly audio: GeneratedAudioFile;
     /**
-  The ID of the response model that was used to generate the response.
+    Warnings for the call, e.g. unsupported settings.
+       */
+    readonly warnings: Array<SpeechWarning>;
+    /**
+    Response metadata from the provider. There may be multiple responses if we made multiple calls to the model.
      */
-    modelId: string;
+    readonly responses: Array<SpeechModelResponseMetadata>;
     /**
-  Response headers.
+    Provider metadata from the provider.
      */
+    readonly providerMetadata: Record<string, Record<string, JSONValue>>;
+}
+
+/**
+Generates speech audio using a speech model.
+
+@param model - The speech model to use.
+@param text - The text to convert to speech.
+@param voice - The voice to use for speech generation.
+@param outputFormat - The output format to use for speech generation e.g. "mp3", "wav", etc.
+@param instructions - Instructions for the speech generation e.g. "Speak in a slow and steady tone".
+@param speed - The speed of the speech generation.
+@param providerOptions - Additional provider-specific options that are passed through to the provider
+as body parameters.
+@param maxRetries - Maximum number of retries. Set to 0 to disable retries. Default: 2.
+@param abortSignal - An optional abort signal that can be used to cancel the call.
+@param headers - Additional HTTP headers to be sent with the request. Only applicable for HTTP-based providers.
+
+@returns A result object that contains the generated audio data.
+ */
+declare function generateSpeech({ model, text, voice, outputFormat, instructions, speed, providerOptions, maxRetries: maxRetriesArg, abortSignal, headers, }: {
+    /**
+  The speech model to use.
+       */
+    model: SpeechModelV1;
+    /**
+  The text to convert to speech.
+     */
+    text: string;
+    /**
+  The voice to use for speech generation.
+     */
+    voice?: string;
+    /**
+     * The desired output format for the audio e.g. "mp3", "wav", etc.
+     */
+    outputFormat?: 'mp3' | 'wav' | (string & {});
+    /**
+      Instructions for the speech generation e.g. "Speak in a slow and steady tone".
+    */
+    instructions?: string;
+    /**
+    The speed of the speech generation.
+     */
+    speed?: number;
+    /**
+  Additional provider-specific options that are passed through to the provider
+  as body parameters.
+  
+  The outer record is keyed by the provider name, and the inner
+  record is keyed by the provider-specific metadata key.
+  ```ts
+  {
+    "openai": {}
+  }
+  ```
+       */
+    providerOptions?: ProviderOptions;
+    /**
+  Maximum number of retries per speech model call. Set to 0 to disable retries.
+  
+  @default 2
+     */
+    maxRetries?: number;
+    /**
+  Abort signal.
+   */
+    abortSignal?: AbortSignal;
+    /**
+  Additional headers to include in the request.
+  Only applicable for HTTP-based providers.
+   */
     headers?: Record<string, string>;
-};
+}): Promise<SpeechResult>;
 
 /**
 The result of a `transcribe` call.
@@ -4521,4 +4655,4 @@ declare namespace llamaindexAdapter {
   };
 }
 
-export { AssistantContent, AssistantResponse, CallWarning, ChunkDetector, CoreAssistantMessage, CoreMessage, CoreSystemMessage, CoreTool, CoreToolCallUnion, CoreToolChoice, CoreToolMessage, CoreToolResultUnion, CoreUserMessage, DataContent, DataStreamOptions, DataStreamWriter, DownloadError, EmbedManyResult, EmbedResult, Embedding, EmbeddingModel, EmbeddingModelUsage, GenerateImageResult as Experimental_GenerateImageResult, GeneratedFile as Experimental_GeneratedImage, Experimental_LanguageModelV1Middleware, TranscriptionResult as Experimental_TranscriptionResult, FilePart, FinishReason, GenerateObjectResult, GenerateTextOnStepFinishCallback, GenerateTextResult, GeneratedFile, ImageModel, ImageGenerationWarning as ImageModelCallWarning, ImageModelResponseMetadata, ImagePart, InvalidArgumentError, InvalidDataContentError, InvalidMessageRoleError, InvalidStreamPartError, InvalidToolArgumentsError, JSONRPCError, JSONRPCMessage, JSONRPCNotification, JSONRPCRequest, JSONRPCResponse, langchainAdapter as LangChainAdapter, LanguageModel, LanguageModelRequestMetadata, LanguageModelResponseMetadata, LanguageModelUsage, LanguageModelV1Middleware, llamaindexAdapter as LlamaIndexAdapter, LogProbs, MCPClientError, MCPTransport, MessageConversionError, NoImageGeneratedError, NoObjectGeneratedError, NoOutputSpecifiedError, NoSuchProviderError, NoSuchToolError, ObjectStreamPart, output as Output, Provider, ProviderMetadata, ProviderRegistryProvider, RepairTextFunction, RetryError, StepResult, StreamData, StreamObjectOnFinishCallback, StreamObjectResult, StreamTextOnChunkCallback, StreamTextOnErrorCallback, StreamTextOnFinishCallback, StreamTextOnStepFinishCallback, StreamTextResult, StreamTextTransform, TelemetrySettings, TextPart, TextStreamPart, Tool, ToolCallPart, ToolCallRepairError, ToolCallRepairFunction, ToolCallUnion, ToolChoice, ToolContent, ToolExecutionError, ToolExecutionOptions, ToolResultPart, ToolResultUnion, ToolSet, UserContent, appendClientMessage, appendResponseMessages, convertToCoreMessages, coreAssistantMessageSchema, coreMessageSchema, coreSystemMessageSchema, coreToolMessageSchema, coreUserMessageSchema, cosineSimilarity, createDataStream, createDataStreamResponse, createProviderRegistry, customProvider, defaultSettingsMiddleware, embed, embedMany, createMCPClient as experimental_createMCPClient, experimental_createProviderRegistry, experimental_customProvider, generateImage as experimental_generateImage, transcribe as experimental_transcribe, experimental_wrapLanguageModel, extractReasoningMiddleware, generateObject, generateText, pipeDataStreamToResponse, simulateReadableStream, simulateStreamingMiddleware, smoothStream, streamObject, streamText, tool, wrapLanguageModel };
+export { AssistantContent, AssistantResponse, CallWarning, ChunkDetector, CoreAssistantMessage, CoreMessage, CoreSystemMessage, CoreTool, CoreToolCallUnion, CoreToolChoice, CoreToolMessage, CoreToolResultUnion, CoreUserMessage, DataContent, DataStreamOptions, DataStreamWriter, DownloadError, EmbedManyResult, EmbedResult, Embedding, EmbeddingModel, EmbeddingModelUsage, GenerateImageResult as Experimental_GenerateImageResult, GeneratedFile as Experimental_GeneratedImage, Experimental_LanguageModelV1Middleware, SpeechResult as Experimental_SpeechResult, TranscriptionResult as Experimental_TranscriptionResult, FilePart, FinishReason, GenerateObjectResult, GenerateTextOnStepFinishCallback, GenerateTextResult, GeneratedAudioFile, GeneratedFile, ImageModel, ImageGenerationWarning as ImageModelCallWarning, ImageModelResponseMetadata, ImagePart, InvalidArgumentError, InvalidDataContentError, InvalidMessageRoleError, InvalidStreamPartError, InvalidToolArgumentsError, JSONRPCError, JSONRPCMessage, JSONRPCNotification, JSONRPCRequest, JSONRPCResponse, langchainAdapter as LangChainAdapter, LanguageModel, LanguageModelRequestMetadata, LanguageModelResponseMetadata, LanguageModelUsage, LanguageModelV1Middleware, llamaindexAdapter as LlamaIndexAdapter, LogProbs, MCPClientError, MCPTransport, MessageConversionError, NoImageGeneratedError, NoObjectGeneratedError, NoOutputSpecifiedError, NoSuchProviderError, NoSuchToolError, ObjectStreamPart, output as Output, Provider, ProviderMetadata, ProviderRegistryProvider, RepairTextFunction, RetryError, SpeechModel, SpeechModelResponseMetadata, SpeechWarning, StepResult, StreamData, StreamObjectOnFinishCallback, StreamObjectResult, StreamTextOnChunkCallback, StreamTextOnErrorCallback, StreamTextOnFinishCallback, StreamTextOnStepFinishCallback, StreamTextResult, StreamTextTransform, TelemetrySettings, TextPart, TextStreamPart, Tool, ToolCallPart, ToolCallRepairError, ToolCallRepairFunction, ToolCallUnion, ToolChoice, ToolContent, ToolExecutionError, ToolExecutionOptions, ToolResultPart, ToolResultUnion, ToolSet, TranscriptionModel, TranscriptionModelResponseMetadata, TranscriptionWarning, UserContent, appendClientMessage, appendResponseMessages, convertToCoreMessages, coreAssistantMessageSchema, coreMessageSchema, coreSystemMessageSchema, coreToolMessageSchema, coreUserMessageSchema, cosineSimilarity, createDataStream, createDataStreamResponse, createProviderRegistry, customProvider, defaultSettingsMiddleware, embed, embedMany, createMCPClient as experimental_createMCPClient, experimental_createProviderRegistry, experimental_customProvider, generateImage as experimental_generateImage, generateSpeech as experimental_generateSpeech, transcribe as experimental_transcribe, experimental_wrapLanguageModel, extractReasoningMiddleware, generateObject, generateText, pipeDataStreamToResponse, simulateReadableStream, simulateStreamingMiddleware, smoothStream, streamObject, streamText, tool, wrapLanguageModel };

package/dist/index.js

@@ -73,6 +73,7 @@ __export(streams_exports, {
   experimental_createProviderRegistry: () => experimental_createProviderRegistry,
   experimental_customProvider: () => experimental_customProvider,
   experimental_generateImage: () => generateImage,
+  experimental_generateSpeech: () => generateSpeech,
   experimental_transcribe: () => transcribe,
   experimental_wrapLanguageModel: () => experimental_wrapLanguageModel,
   extractReasoningMiddleware: () => extractReasoningMiddleware,
@@ -6335,9 +6336,99 @@ var DefaultStreamTextResult = class {
   }
 };
 
-// errors/no-transcript-generated-error.ts
+// errors/no-speech-generated-error.ts
 var import_provider23 = require("@ai-sdk/provider");
-var NoTranscriptGeneratedError = class extends import_provider23.AISDKError {
+var NoSpeechGeneratedError = class extends import_provider23.AISDKError {
+  constructor(options) {
+    super({
+      name: "AI_NoSpeechGeneratedError",
+      message: "No speech audio generated."
+    });
+    this.responses = options.responses;
+  }
+};
+
+// core/generate-speech/generated-audio-file.ts
+var DefaultGeneratedAudioFile = class extends DefaultGeneratedFile {
+  constructor({
+    data,
+    mimeType
+  }) {
+    super({ data, mimeType });
+    let format = "mp3";
+    if (mimeType) {
+      const mimeTypeParts = mimeType.split("/");
+      if (mimeTypeParts.length === 2) {
+        if (mimeType !== "audio/mpeg") {
+          format = mimeTypeParts[1];
+        }
+      }
+    }
+    if (!format) {
+      throw new Error(
+        "Audio format must be provided or determinable from mimeType"
+      );
+    }
+    this.format = format;
+  }
+};
+
+// core/generate-speech/generate-speech.ts
+async function generateSpeech({
+  model,
+  text: text2,
+  voice,
+  outputFormat,
+  instructions,
+  speed,
+  providerOptions = {},
+  maxRetries: maxRetriesArg,
+  abortSignal,
+  headers
+}) {
+  var _a17;
+  const { retry } = prepareRetries({ maxRetries: maxRetriesArg });
+  const result = await retry(
+    () => model.doGenerate({
+      text: text2,
+      voice,
+      outputFormat,
+      instructions,
+      speed,
+      abortSignal,
+      headers,
+      providerOptions
+    })
+  );
+  if (!result.audio || result.audio.length === 0) {
+    throw new NoSpeechGeneratedError({ responses: [result.response] });
+  }
+  return new DefaultSpeechResult({
+    audio: new DefaultGeneratedAudioFile({
+      data: result.audio,
+      mimeType: (_a17 = detectMimeType({
+        data: result.audio,
+        signatures: audioMimeTypeSignatures
+      })) != null ? _a17 : "audio/mp3"
+    }),
+    warnings: result.warnings,
+    responses: [result.response],
+    providerMetadata: result.providerMetadata
+  });
+}
+var DefaultSpeechResult = class {
+  constructor(options) {
+    var _a17;
+    this.audio = options.audio;
+    this.warnings = options.warnings;
+    this.responses = options.responses;
+    this.providerMetadata = (_a17 = options.providerMetadata) != null ? _a17 : {};
+  }
+};
+
+// errors/no-transcript-generated-error.ts
+var import_provider24 = require("@ai-sdk/provider");
+var NoTranscriptGeneratedError = class extends import_provider24.AISDKError {
   constructor(options) {
     super({
       name: "AI_NoTranscriptGeneratedError",
@@ -6714,7 +6805,7 @@ function appendClientMessage({
 
 // core/prompt/append-response-messages.ts
 var import_ui_utils9 = require("@ai-sdk/ui-utils");
-var import_provider24 = require("@ai-sdk/provider");
+var import_provider25 = require("@ai-sdk/provider");
 function appendResponseMessages({
   messages,
   responseMessages,
@@ -6797,7 +6888,7 @@ function appendResponseMessages({
                 break;
               case "file":
                 if (part.data instanceof URL) {
-                  throw new import_provider24.AISDKError({
+                  throw new import_provider25.AISDKError({
                     name: "InvalidAssistantFileData",
                     message: "File data cannot be a URL"
                   });
@@ -6891,7 +6982,7 @@ function appendResponseMessages({
 }
 
 // core/registry/custom-provider.ts
-var import_provider25 = require("@ai-sdk/provider");
+var import_provider26 = require("@ai-sdk/provider");
 function customProvider({
   languageModels,
   textEmbeddingModels,
@@ -6906,7 +6997,7 @@ function customProvider({
       if (fallbackProvider) {
         return fallbackProvider.languageModel(modelId);
       }
-      throw new import_provider25.NoSuchModelError({ modelId, modelType: "languageModel" });
+      throw new import_provider26.NoSuchModelError({ modelId, modelType: "languageModel" });
     },
     textEmbeddingModel(modelId) {
       if (textEmbeddingModels != null && modelId in textEmbeddingModels) {
@@ -6915,7 +7006,7 @@ function customProvider({
       if (fallbackProvider) {
         return fallbackProvider.textEmbeddingModel(modelId);
       }
-      throw new import_provider25.NoSuchModelError({ modelId, modelType: "textEmbeddingModel" });
+      throw new import_provider26.NoSuchModelError({ modelId, modelType: "textEmbeddingModel" });
     },
     imageModel(modelId) {
       if (imageModels != null && modelId in imageModels) {
@@ -6924,19 +7015,19 @@ function customProvider({
       if (fallbackProvider == null ? void 0 : fallbackProvider.imageModel) {
         return fallbackProvider.imageModel(modelId);
       }
-      throw new import_provider25.NoSuchModelError({ modelId, modelType: "imageModel" });
+      throw new import_provider26.NoSuchModelError({ modelId, modelType: "imageModel" });
     }
   };
 }
 var experimental_customProvider = customProvider;
 
 // core/registry/no-such-provider-error.ts
-var import_provider26 = require("@ai-sdk/provider");
+var import_provider27 = require("@ai-sdk/provider");
 var name16 = "AI_NoSuchProviderError";
 var marker16 = `vercel.ai.error.${name16}`;
 var symbol16 = Symbol.for(marker16);
 var _a16;
-var NoSuchProviderError = class extends import_provider26.NoSuchModelError {
+var NoSuchProviderError = class extends import_provider27.NoSuchModelError {
   constructor({
     modelId,
     modelType,
@@ -6950,13 +7041,13 @@ var NoSuchProviderError = class extends import_provider26.NoSuchModelError {
     this.availableProviders = availableProviders;
   }
   static isInstance(error) {
-    return import_provider26.AISDKError.hasMarker(error, marker16);
+    return import_provider27.AISDKError.hasMarker(error, marker16);
   }
 };
 _a16 = symbol16;
 
 // core/registry/provider-registry.ts
-var import_provider27 = require("@ai-sdk/provider");
+var import_provider28 = require("@ai-sdk/provider");
 function createProviderRegistry(providers, {
   separator = ":"
 } = {}) {
@@ -6995,20 +7086,20 @@ var DefaultProviderRegistry = class {
   splitId(id, modelType) {
     const index = id.indexOf(this.separator);
     if (index === -1) {
-      throw new import_provider27.NoSuchModelError({
+      throw new import_provider28.NoSuchModelError({
         modelId: id,
         modelType,
         message: `Invalid ${modelType} id for registry: ${id} (must be in the format "providerId${this.separator}modelId")`
       });
     }
-    return [id.slice(0, index), id.slice(index + 1)];
+    return [id.slice(0, index), id.slice(index + this.separator.length)];
   }
   languageModel(id) {
     var _a17, _b;
     const [providerId, modelId] = this.splitId(id, "languageModel");
     const model = (_b = (_a17 = this.getProvider(providerId)).languageModel) == null ? void 0 : _b.call(_a17, modelId);
     if (model == null) {
-      throw new import_provider27.NoSuchModelError({ modelId: id, modelType: "languageModel" });
+      throw new import_provider28.NoSuchModelError({ modelId: id, modelType: "languageModel" });
     }
     return model;
   }
@@ -7018,7 +7109,7 @@ var DefaultProviderRegistry = class {
     const provider = this.getProvider(providerId);
     const model = (_a17 = provider.textEmbeddingModel) == null ? void 0 : _a17.call(provider, modelId);
     if (model == null) {
-      throw new import_provider27.NoSuchModelError({
+      throw new import_provider28.NoSuchModelError({
         modelId: id,
         modelType: "textEmbeddingModel"
       });
@@ -7031,7 +7122,7 @@ var DefaultProviderRegistry = class {
     const provider = this.getProvider(providerId);
     const model = (_a17 = provider.imageModel) == null ? void 0 : _a17.call(provider, modelId);
     if (model == null) {
-      throw new import_provider27.NoSuchModelError({ modelId: id, modelType: "imageModel" });
+      throw new import_provider28.NoSuchModelError({ modelId: id, modelType: "imageModel" });
     }
     return model;
   }
@@ -8038,6 +8129,7 @@ var StreamData = class {
   experimental_createProviderRegistry,
   experimental_customProvider,
   experimental_generateImage,
+  experimental_generateSpeech,
   experimental_transcribe,
   experimental_wrapLanguageModel,
   extractReasoningMiddleware,