ai 4.3.2 → 4.3.4

package/CHANGELOG.md

@@ -1,5 +1,28 @@
 # ai
 
+## 4.3.4
+
+### Patch Changes
+
+- 013faa8: core (ai): change transcription model mimeType to mediaType
+- Updated dependencies [013faa8]
+  - @ai-sdk/provider@1.1.2
+  - @ai-sdk/provider-utils@2.2.6
+  - @ai-sdk/ui-utils@1.2.7
+  - @ai-sdk/react@1.2.8
+
+## 4.3.3
+
+### Patch Changes
+
+- 3e88f4d: fix (ai/mcp): prevent mutation of customEnv
+- c21fa6d: feat: add transcription with experimental_transcribe
+- Updated dependencies [c21fa6d]
+  - @ai-sdk/provider-utils@2.2.5
+  - @ai-sdk/provider@1.1.1
+  - @ai-sdk/react@1.2.7
+  - @ai-sdk/ui-utils@1.2.6
+
 ## 4.3.2
 
 ### 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, ProviderV1, NoSuchModelError } from '@ai-sdk/provider';
+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';
 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';
@@ -3792,6 +3792,132 @@ 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;
+
+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>;
+};
+
+/**
+The result of a `transcribe` call.
+It contains the transcript and additional information.
+ */
+interface TranscriptionResult {
+    /**
+     * The complete transcribed text from the audio.
+     */
+    readonly text: string;
+    /**
+     * Array of transcript segments with timing information.
+     * Each segment represents a portion of the transcribed text with start and end times.
+     */
+    readonly segments: Array<{
+        /**
+         * The text content of this segment.
+         */
+        readonly text: string;
+        /**
+         * The start time of this segment in seconds.
+         */
+        readonly startSecond: number;
+        /**
+         * The end time of this segment in seconds.
+         */
+        readonly endSecond: number;
+    }>;
+    /**
+     * The detected language of the audio content, as an ISO-639-1 code (e.g., 'en' for English).
+     * May be undefined if the language couldn't be detected.
+     */
+    readonly language: string | undefined;
+    /**
+     * The total duration of the audio file in seconds.
+     * May be undefined if the duration couldn't be determined.
+     */
+    readonly durationInSeconds: number | undefined;
+    /**
+    Warnings for the call, e.g. unsupported settings.
+       */
+    readonly warnings: Array<TranscriptionWarning>;
+    /**
+    Response metadata from the provider. There may be multiple responses if we made multiple calls to the model.
+     */
+    readonly responses: Array<TranscriptionModelResponseMetadata>;
+    /**
+    Provider metadata from the provider.
+     */
+    readonly providerMetadata: Record<string, Record<string, JSONValue>>;
+}
+
+/**
+Generates transcripts using a transcription model.
+
+@param model - The transcription model to use.
+@param audio - The audio data to transcribe as DataContent (string | Uint8Array | ArrayBuffer | Buffer) or a URL.
+@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 transcript.
+ */
+declare function transcribe({ model, audio, providerOptions, maxRetries: maxRetriesArg, abortSignal, headers, }: {
+    /**
+  The transcription model to use.
+       */
+    model: TranscriptionModelV1;
+    /**
+  The audio data to transcribe.
+     */
+    audio: DataContent | URL;
+    /**
+  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": {
+      "temperature": 0
+    }
+  }
+  ```
+       */
+    providerOptions?: ProviderOptions;
+    /**
+  Maximum number of retries per transcript 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<TranscriptionResult>;
+
 /**
  * Experimental middleware for LanguageModelV1.
  * This type defines the structure for middleware that can be used to modify
@@ -4395,4 +4521,4 @@ 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, 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, 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, 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 };

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, ProviderV1, NoSuchModelError } from '@ai-sdk/provider';
+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';
 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';
@@ -3792,6 +3792,132 @@ 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;
+
+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>;
+};
+
+/**
+The result of a `transcribe` call.
+It contains the transcript and additional information.
+ */
+interface TranscriptionResult {
+    /**
+     * The complete transcribed text from the audio.
+     */
+    readonly text: string;
+    /**
+     * Array of transcript segments with timing information.
+     * Each segment represents a portion of the transcribed text with start and end times.
+     */
+    readonly segments: Array<{
+        /**
+         * The text content of this segment.
+         */
+        readonly text: string;
+        /**
+         * The start time of this segment in seconds.
+         */
+        readonly startSecond: number;
+        /**
+         * The end time of this segment in seconds.
+         */
+        readonly endSecond: number;
+    }>;
+    /**
+     * The detected language of the audio content, as an ISO-639-1 code (e.g., 'en' for English).
+     * May be undefined if the language couldn't be detected.
+     */
+    readonly language: string | undefined;
+    /**
+     * The total duration of the audio file in seconds.
+     * May be undefined if the duration couldn't be determined.
+     */
+    readonly durationInSeconds: number | undefined;
+    /**
+    Warnings for the call, e.g. unsupported settings.
+       */
+    readonly warnings: Array<TranscriptionWarning>;
+    /**
+    Response metadata from the provider. There may be multiple responses if we made multiple calls to the model.
+     */
+    readonly responses: Array<TranscriptionModelResponseMetadata>;
+    /**
+    Provider metadata from the provider.
+     */
+    readonly providerMetadata: Record<string, Record<string, JSONValue>>;
+}
+
+/**
+Generates transcripts using a transcription model.
+
+@param model - The transcription model to use.
+@param audio - The audio data to transcribe as DataContent (string | Uint8Array | ArrayBuffer | Buffer) or a URL.
+@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 transcript.
+ */
+declare function transcribe({ model, audio, providerOptions, maxRetries: maxRetriesArg, abortSignal, headers, }: {
+    /**
+  The transcription model to use.
+       */
+    model: TranscriptionModelV1;
+    /**
+  The audio data to transcribe.
+     */
+    audio: DataContent | URL;
+    /**
+  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": {
+      "temperature": 0
+    }
+  }
+  ```
+       */
+    providerOptions?: ProviderOptions;
+    /**
+  Maximum number of retries per transcript 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<TranscriptionResult>;
+
 /**
  * Experimental middleware for LanguageModelV1.
  * This type defines the structure for middleware that can be used to modify
@@ -4395,4 +4521,4 @@ 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, 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, 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, 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 };

package/dist/index.js

@@ -73,6 +73,7 @@ __export(streams_exports, {
   experimental_createProviderRegistry: () => experimental_createProviderRegistry,
   experimental_customProvider: () => experimental_customProvider,
   experimental_generateImage: () => generateImage,
+  experimental_transcribe: () => transcribe,
   experimental_wrapLanguageModel: () => experimental_wrapLanguageModel,
   extractReasoningMiddleware: () => extractReasoningMiddleware,
   formatAssistantStreamPart: () => import_ui_utils11.formatAssistantStreamPart,
@@ -942,8 +943,8 @@ var DefaultGeneratedFileWithType = class extends DefaultGeneratedFile {
   }
 };
 
-// core/util/detect-image-mimetype.ts
-var mimeTypeSignatures = [
+// core/util/detect-mimetype.ts
+var imageMimeTypeSignatures = [
   {
     mimeType: "image/gif",
     bytesPrefix: [71, 73, 70],
@@ -1016,9 +1017,44 @@ var mimeTypeSignatures = [
     base64Prefix: "AAAAIGZ0eXBoZWlj"
   }
 ];
-function detectImageMimeType(image) {
-  for (const signature of mimeTypeSignatures) {
-    if (typeof image === "string" ? image.startsWith(signature.base64Prefix) : image.length >= signature.bytesPrefix.length && signature.bytesPrefix.every((byte, index) => image[index] === byte)) {
+var audioMimeTypeSignatures = [
+  {
+    mimeType: "audio/mpeg",
+    bytesPrefix: [255, 251],
+    base64Prefix: "//s="
+  },
+  {
+    mimeType: "audio/wav",
+    bytesPrefix: [82, 73, 70, 70],
+    base64Prefix: "UklGR"
+  },
+  {
+    mimeType: "audio/ogg",
+    bytesPrefix: [79, 103, 103, 83],
+    base64Prefix: "T2dnUw"
+  },
+  {
+    mimeType: "audio/flac",
+    bytesPrefix: [102, 76, 97, 67],
+    base64Prefix: "ZkxhQw"
+  },
+  {
+    mimeType: "audio/aac",
+    bytesPrefix: [64, 21, 0, 0],
+    base64Prefix: "QBUA"
+  },
+  {
+    mimeType: "audio/mp4",
+    bytesPrefix: [102, 116, 121, 112],
+    base64Prefix: "ZnR5cA"
+  }
+];
+function detectMimeType({
+  data,
+  signatures
+}) {
+  for (const signature of signatures) {
+    if (typeof data === "string" ? data.startsWith(signature.base64Prefix) : data.length >= signature.bytesPrefix.length && signature.bytesPrefix.every((byte, index) => data[index] === byte)) {
       return signature.mimeType;
     }
   }
@@ -1075,7 +1111,10 @@ async function generateImage({
           var _a18;
           return new DefaultGeneratedFile({
             data: image,
-            mimeType: (_a18 = detectImageMimeType(image)) != null ? _a18 : "image/png"
+            mimeType: (_a18 = detectMimeType({
+              data: image,
+              signatures: imageMimeTypeSignatures
+            })) != null ? _a18 : "image/png"
           });
         }
       )
@@ -1504,7 +1543,10 @@ function convertPartToLanguageModelPart(part, downloadedAssets) {
   switch (type) {
     case "image": {
       if (normalizedData instanceof Uint8Array) {
-        mimeType = (_b = detectImageMimeType(normalizedData)) != null ? _b : mimeType;
+        mimeType = (_b = detectMimeType({
+          data: normalizedData,
+          signatures: imageMimeTypeSignatures
+        })) != null ? _b : mimeType;
       }
       return {
         type: "image",
@@ -6293,6 +6335,70 @@ var DefaultStreamTextResult = class {
   }
 };
 
+// errors/no-transcript-generated-error.ts
+var import_provider23 = require("@ai-sdk/provider");
+var NoTranscriptGeneratedError = class extends import_provider23.AISDKError {
+  constructor(options) {
+    super({
+      name: "AI_NoTranscriptGeneratedError",
+      message: "No transcript generated."
+    });
+    this.responses = options.responses;
+  }
+};
+
+// core/transcribe/transcribe.ts
+async function transcribe({
+  model,
+  audio,
+  providerOptions = {},
+  maxRetries: maxRetriesArg,
+  abortSignal,
+  headers
+}) {
+  const { retry } = prepareRetries({ maxRetries: maxRetriesArg });
+  const audioData = audio instanceof URL ? (await download({ url: audio })).data : convertDataContentToUint8Array(audio);
+  const result = await retry(
+    () => {
+      var _a17;
+      return model.doGenerate({
+        audio: audioData,
+        abortSignal,
+        headers,
+        providerOptions,
+        mediaType: (_a17 = detectMimeType({
+          data: audioData,
+          signatures: audioMimeTypeSignatures
+        })) != null ? _a17 : "audio/wav"
+      });
+    }
+  );
+  if (!result.text) {
+    throw new NoTranscriptGeneratedError({ responses: [result.response] });
+  }
+  return new DefaultTranscriptionResult({
+    text: result.text,
+    segments: result.segments,
+    language: result.language,
+    durationInSeconds: result.durationInSeconds,
+    warnings: result.warnings,
+    responses: [result.response],
+    providerMetadata: result.providerMetadata
+  });
+}
+var DefaultTranscriptionResult = class {
+  constructor(options) {
+    var _a17;
+    this.text = options.text;
+    this.segments = options.segments;
+    this.language = options.language;
+    this.durationInSeconds = options.durationInSeconds;
+    this.warnings = options.warnings;
+    this.responses = options.responses;
+    this.providerMetadata = (_a17 = options.providerMetadata) != null ? _a17 : {};
+  }
+};
+
 // core/util/merge-objects.ts
 function mergeObjects(target, source) {
   if (target === void 0 && source === void 0) {
@@ -6608,7 +6714,7 @@ function appendClientMessage({
 
 // core/prompt/append-response-messages.ts
 var import_ui_utils9 = require("@ai-sdk/ui-utils");
-var import_provider23 = require("@ai-sdk/provider");
+var import_provider24 = require("@ai-sdk/provider");
 function appendResponseMessages({
   messages,
   responseMessages,
@@ -6691,7 +6797,7 @@ function appendResponseMessages({
                 break;
               case "file":
                 if (part.data instanceof URL) {
-                  throw new import_provider23.AISDKError({
+                  throw new import_provider24.AISDKError({
                     name: "InvalidAssistantFileData",
                     message: "File data cannot be a URL"
                   });
@@ -6785,7 +6891,7 @@ function appendResponseMessages({
 }
 
 // core/registry/custom-provider.ts
-var import_provider24 = require("@ai-sdk/provider");
+var import_provider25 = require("@ai-sdk/provider");
 function customProvider({
   languageModels,
   textEmbeddingModels,
@@ -6800,7 +6906,7 @@ function customProvider({
       if (fallbackProvider) {
         return fallbackProvider.languageModel(modelId);
       }
-      throw new import_provider24.NoSuchModelError({ modelId, modelType: "languageModel" });
+      throw new import_provider25.NoSuchModelError({ modelId, modelType: "languageModel" });
     },
     textEmbeddingModel(modelId) {
       if (textEmbeddingModels != null && modelId in textEmbeddingModels) {
@@ -6809,7 +6915,7 @@ function customProvider({
       if (fallbackProvider) {
         return fallbackProvider.textEmbeddingModel(modelId);
       }
-      throw new import_provider24.NoSuchModelError({ modelId, modelType: "textEmbeddingModel" });
+      throw new import_provider25.NoSuchModelError({ modelId, modelType: "textEmbeddingModel" });
     },
     imageModel(modelId) {
       if (imageModels != null && modelId in imageModels) {
@@ -6818,19 +6924,19 @@ function customProvider({
       if (fallbackProvider == null ? void 0 : fallbackProvider.imageModel) {
         return fallbackProvider.imageModel(modelId);
       }
-      throw new import_provider24.NoSuchModelError({ modelId, modelType: "imageModel" });
+      throw new import_provider25.NoSuchModelError({ modelId, modelType: "imageModel" });
     }
   };
 }
 var experimental_customProvider = customProvider;
 
 // core/registry/no-such-provider-error.ts
-var import_provider25 = require("@ai-sdk/provider");
+var import_provider26 = 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_provider25.NoSuchModelError {
+var NoSuchProviderError = class extends import_provider26.NoSuchModelError {
   constructor({
     modelId,
     modelType,
@@ -6844,13 +6950,13 @@ var NoSuchProviderError = class extends import_provider25.NoSuchModelError {
     this.availableProviders = availableProviders;
   }
   static isInstance(error) {
-    return import_provider25.AISDKError.hasMarker(error, marker16);
+    return import_provider26.AISDKError.hasMarker(error, marker16);
   }
 };
 _a16 = symbol16;
 
 // core/registry/provider-registry.ts
-var import_provider26 = require("@ai-sdk/provider");
+var import_provider27 = require("@ai-sdk/provider");
 function createProviderRegistry(providers, {
   separator = ":"
 } = {}) {
@@ -6889,7 +6995,7 @@ var DefaultProviderRegistry = class {
   splitId(id, modelType) {
     const index = id.indexOf(this.separator);
     if (index === -1) {
-      throw new import_provider26.NoSuchModelError({
+      throw new import_provider27.NoSuchModelError({
         modelId: id,
         modelType,
         message: `Invalid ${modelType} id for registry: ${id} (must be in the format "providerId${this.separator}modelId")`
@@ -6902,7 +7008,7 @@ var DefaultProviderRegistry = class {
     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_provider26.NoSuchModelError({ modelId: id, modelType: "languageModel" });
+      throw new import_provider27.NoSuchModelError({ modelId: id, modelType: "languageModel" });
     }
     return model;
   }
@@ -6912,7 +7018,7 @@ 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_provider26.NoSuchModelError({
+      throw new import_provider27.NoSuchModelError({
         modelId: id,
         modelType: "textEmbeddingModel"
       });
@@ -6925,7 +7031,7 @@ 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_provider26.NoSuchModelError({ modelId: id, modelType: "imageModel" });
+      throw new import_provider27.NoSuchModelError({ modelId: id, modelType: "imageModel" });
     }
     return model;
   }
@@ -7927,6 +8033,7 @@ var StreamData = class {
   experimental_createProviderRegistry,
   experimental_customProvider,
   experimental_generateImage,
+  experimental_transcribe,
   experimental_wrapLanguageModel,
   extractReasoningMiddleware,
   formatAssistantStreamPart,