Backport: feat(deepgram): add text-to-speech support (#10429)

This is an automated backport of #10063 to the release-v5.0 branch.

---------

Co-authored-by: dpmishler <122304480+dpmishler@users.noreply.github.com>
Co-authored-by: Gregor Martynus <39992+gr2m@users.noreply.github.com>

Author: 3 people

.changeset/feat-deepgram-tts.md

@@ -0,0 +1,7 @@
+---
+'@ai-sdk/deepgram': patch
+---
+
+feat(deepgram): add text-to-speech support
+
+Add text-to-speech support for Deepgram Aura models via the REST API. Supports all Aura voice models (aura-2-helena-en, aura-2-thalia-en, etc.) with proper audio format validation, encoding/container/sample_rate/bitrate combinations, and comprehensive parameter validation.

examples/ai-core/src/generate-speech/deepgram.ts

@@ -0,0 +1,20 @@
+import { deepgram } from '@ai-sdk/deepgram';
+import { experimental_generateSpeech as generateSpeech } from 'ai';
+import 'dotenv/config';
+import { saveAudioFile } from '../lib/save-audio';
+
+async function main() {
+  const result = await generateSpeech({
+    model: deepgram.speech('aura-2-helena-en'),
+    text: 'Hello, welcome to Deepgram! This is a test of the text-to-speech API.',
+  });
+
+  console.log('Audio:', result.audio);
+  console.log('Warnings:', result.warnings);
+  console.log('Responses:', result.responses);
+  console.log('Provider Metadata:', result.providerMetadata);
+
+  await saveAudioFile(result.audio);
+}
+
+main().catch(console.error);

packages/deepgram/README.md

@@ -1,7 +1,7 @@
 # AI SDK - Deepgram Provider
 
 The **[Deepgram provider](https://ai-sdk.dev/providers/ai-sdk-providers/deepgram)** for the [AI SDK](https://ai-sdk.dev/docs)
-contains transcription model support for the Deepgram transcription API.
+contains transcription model support for the Deepgram transcription API and speech model support for the Deepgram text-to-speech API.
 
 ## Setup
 
@@ -19,7 +19,9 @@ You can import the default provider instance `deepgram` from `@ai-sdk/deepgram`:
 import { deepgram } from '@ai-sdk/deepgram';
 ```
 
-## Example
+## Examples
+
+### Transcription
 
 ```ts
 import { deepgram } from '@ai-sdk/deepgram';
@@ -33,6 +35,18 @@ const { text } = await transcribe({
 });
 ```
 
+### Text-to-Speech
+
+```ts
+import { deepgram } from '@ai-sdk/deepgram';
+import { experimental_generateSpeech as generateSpeech } from 'ai';
+
+const { audio } = await generateSpeech({
+  model: deepgram.speech('aura-2-helena-en'),
+  text: 'Hello, welcome to Deepgram!',
+});
+```
+
 ## Documentation
 
 Please check out the **[Deepgram provider documentation](https://ai-sdk.dev/providers/ai-sdk-providers/deepgram)** for more information.

packages/deepgram/src/deepgram-provider.ts

@@ -1,5 +1,6 @@
 import {
   TranscriptionModelV2,
+  SpeechModelV2,
   ProviderV2,
   NoSuchModelError,
 } from '@ai-sdk/provider';
@@ -10,6 +11,8 @@ import {
 } from '@ai-sdk/provider-utils';
 import { DeepgramTranscriptionModel } from './deepgram-transcription-model';
 import { DeepgramTranscriptionModelId } from './deepgram-transcription-options';
+import { DeepgramSpeechModel } from './deepgram-speech-model';
+import { DeepgramSpeechModelId } from './deepgram-speech-options';
 import { VERSION } from './version';
 
 export interface DeepgramProvider extends ProviderV2 {
@@ -24,6 +27,11 @@ export interface DeepgramProvider extends ProviderV2 {
 Creates a model for transcription.
    */
   transcription(modelId: DeepgramTranscriptionModelId): TranscriptionModelV2;
+
+  /**
+Creates a model for speech generation.
+   */
+  speech(modelId: DeepgramSpeechModelId): SpeechModelV2;
 }
 
 export interface DeepgramProviderSettings {
@@ -71,6 +79,14 @@ export function createDeepgram(
       fetch: options.fetch,
     });
 
+  const createSpeechModel = (modelId: DeepgramSpeechModelId) =>
+    new DeepgramSpeechModel(modelId, {
+      provider: `deepgram.speech`,
+      url: ({ path }) => `https://api.deepgram.com${path}`,
+      headers: getHeaders,
+      fetch: options.fetch,
+    });
+
   const provider = function (modelId: DeepgramTranscriptionModelId) {
     return {
       transcription: createTranscriptionModel(modelId),
@@ -79,6 +95,8 @@ export function createDeepgram(
 
   provider.transcription = createTranscriptionModel;
   provider.transcriptionModel = createTranscriptionModel;
+  provider.speech = createSpeechModel;
+  provider.speechModel = createSpeechModel;
 
   // Required ProviderV2 methods that are not supported
   provider.languageModel = () => {

packages/deepgram/src/deepgram-speech-api-types.ts

@@ -0,0 +1,15 @@
+export type DeepgramSpeechAPITypes = {
+  // Request body
+  text: string;
+
+  // Query parameters (these are set via query params, not body)
+  model?: string;
+  encoding?: string;
+  sample_rate?: number;
+  bit_rate?: number | string;
+  container?: string;
+  callback?: string;
+  callback_method?: 'POST' | 'PUT';
+  mip_opt_out?: boolean;
+  tag?: string | string[];
+};