@ai-sdk/lmnt 3.0.0-beta.3 → 3.0.0-beta.31
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/CHANGELOG.md +238 -0
- package/README.md +2 -0
- package/dist/index.d.ts +39 -29
- package/dist/index.js +80 -62
- package/dist/index.js.map +1 -1
- package/package.json +11 -11
- package/src/index.ts +1 -1
- package/src/lmnt-config.ts +2 -2
- package/src/lmnt-provider.ts +34 -5
- package/src/lmnt-speech-model-options.ts +69 -0
- package/src/lmnt-speech-model.ts +29 -80
- package/dist/index.d.mts +0 -89
- package/dist/index.mjs +0 -216
- package/dist/index.mjs.map +0 -1
package/package.json
CHANGED
|
@@ -1,10 +1,10 @@
|
|
|
1
1
|
{
|
|
2
2
|
"name": "@ai-sdk/lmnt",
|
|
3
|
-
"version": "3.0.0-beta.
|
|
3
|
+
"version": "3.0.0-beta.31",
|
|
4
|
+
"type": "module",
|
|
4
5
|
"license": "Apache-2.0",
|
|
5
6
|
"sideEffects": false,
|
|
6
7
|
"main": "./dist/index.js",
|
|
7
|
-
"module": "./dist/index.mjs",
|
|
8
8
|
"types": "./dist/index.d.ts",
|
|
9
9
|
"files": [
|
|
10
10
|
"dist/**/*",
|
|
@@ -24,20 +24,20 @@
|
|
|
24
24
|
"./package.json": "./package.json",
|
|
25
25
|
".": {
|
|
26
26
|
"types": "./dist/index.d.ts",
|
|
27
|
-
"import": "./dist/index.
|
|
28
|
-
"
|
|
27
|
+
"import": "./dist/index.js",
|
|
28
|
+
"default": "./dist/index.js"
|
|
29
29
|
}
|
|
30
30
|
},
|
|
31
31
|
"dependencies": {
|
|
32
|
-
"@ai-sdk/provider": "
|
|
33
|
-
"@ai-sdk/provider
|
|
32
|
+
"@ai-sdk/provider-utils": "5.0.0-beta.30",
|
|
33
|
+
"@ai-sdk/provider": "4.0.0-beta.14"
|
|
34
34
|
},
|
|
35
35
|
"devDependencies": {
|
|
36
36
|
"@types/node": "20.17.24",
|
|
37
37
|
"tsup": "^8",
|
|
38
38
|
"typescript": "5.6.3",
|
|
39
39
|
"zod": "3.25.76",
|
|
40
|
-
"@ai-sdk/test-server": "2.0.0-beta.
|
|
40
|
+
"@ai-sdk/test-server": "2.0.0-beta.3",
|
|
41
41
|
"@vercel/ai-tsconfig": "0.0.0"
|
|
42
42
|
},
|
|
43
43
|
"peerDependencies": {
|
|
@@ -47,12 +47,14 @@
|
|
|
47
47
|
"node": ">=18"
|
|
48
48
|
},
|
|
49
49
|
"publishConfig": {
|
|
50
|
-
"access": "public"
|
|
50
|
+
"access": "public",
|
|
51
|
+
"provenance": true
|
|
51
52
|
},
|
|
52
53
|
"homepage": "https://ai-sdk.dev/docs",
|
|
53
54
|
"repository": {
|
|
54
55
|
"type": "git",
|
|
55
|
-
"url": "
|
|
56
|
+
"url": "https://github.com/vercel/ai",
|
|
57
|
+
"directory": "packages/lmnt"
|
|
56
58
|
},
|
|
57
59
|
"bugs": {
|
|
58
60
|
"url": "https://github.com/vercel/ai/issues"
|
|
@@ -64,9 +66,7 @@
|
|
|
64
66
|
"build": "tsup --tsconfig tsconfig.build.json",
|
|
65
67
|
"build:watch": "tsup --tsconfig tsconfig.build.json --watch",
|
|
66
68
|
"clean": "del-cli dist docs",
|
|
67
|
-
"lint": "eslint \"./**/*.ts*\"",
|
|
68
69
|
"type-check": "tsc --noEmit",
|
|
69
|
-
"prettier-check": "prettier --check \"./**/*.ts*\"",
|
|
70
70
|
"test": "pnpm test:node && pnpm test:edge",
|
|
71
71
|
"test:watch": "vitest --config vitest.node.config.js --watch",
|
|
72
72
|
"test:edge": "vitest --config vitest.edge.config.js --run",
|
package/src/index.ts
CHANGED
|
@@ -1,4 +1,4 @@
|
|
|
1
1
|
export { createLMNT, lmnt } from './lmnt-provider';
|
|
2
2
|
export type { LMNTProvider, LMNTProviderSettings } from './lmnt-provider';
|
|
3
|
-
export type { LMNTSpeechModelOptions } from './lmnt-speech-model';
|
|
3
|
+
export type { LMNTSpeechModelOptions } from './lmnt-speech-model-options';
|
|
4
4
|
export { VERSION } from './version';
|
package/src/lmnt-config.ts
CHANGED
|
@@ -1,9 +1,9 @@
|
|
|
1
|
-
import { FetchFunction } from '@ai-sdk/provider-utils';
|
|
1
|
+
import type { FetchFunction } from '@ai-sdk/provider-utils';
|
|
2
2
|
|
|
3
3
|
export type LMNTConfig = {
|
|
4
4
|
provider: string;
|
|
5
5
|
url: (options: { modelId: string; path: string }) => string;
|
|
6
|
-
headers
|
|
6
|
+
headers?: () => Record<string, string | undefined>;
|
|
7
7
|
fetch?: FetchFunction;
|
|
8
8
|
generateId?: () => string;
|
|
9
9
|
};
|
package/src/lmnt-provider.ts
CHANGED
|
@@ -1,14 +1,18 @@
|
|
|
1
|
-
import { SpeechModelV3, ProviderV3 } from '@ai-sdk/provider';
|
|
2
1
|
import {
|
|
3
|
-
|
|
2
|
+
NoSuchModelError,
|
|
3
|
+
type SpeechModelV4,
|
|
4
|
+
type ProviderV4,
|
|
5
|
+
} from '@ai-sdk/provider';
|
|
6
|
+
import {
|
|
4
7
|
loadApiKey,
|
|
5
8
|
withUserAgentSuffix,
|
|
9
|
+
type FetchFunction,
|
|
6
10
|
} from '@ai-sdk/provider-utils';
|
|
7
11
|
import { LMNTSpeechModel } from './lmnt-speech-model';
|
|
8
|
-
import { LMNTSpeechModelId } from './lmnt-speech-options';
|
|
12
|
+
import type { LMNTSpeechModelId } from './lmnt-speech-options';
|
|
9
13
|
import { VERSION } from './version';
|
|
10
14
|
|
|
11
|
-
export interface LMNTProvider extends
|
|
15
|
+
export interface LMNTProvider extends ProviderV4 {
|
|
12
16
|
(
|
|
13
17
|
modelId: 'aurora',
|
|
14
18
|
settings?: {},
|
|
@@ -19,7 +23,7 @@ export interface LMNTProvider extends Pick<ProviderV3, 'speechModel'> {
|
|
|
19
23
|
/**
|
|
20
24
|
* Creates a model for speech synthesis.
|
|
21
25
|
*/
|
|
22
|
-
speech(modelId: LMNTSpeechModelId):
|
|
26
|
+
speech(modelId: LMNTSpeechModelId): SpeechModelV4;
|
|
23
27
|
}
|
|
24
28
|
|
|
25
29
|
export interface LMNTProviderSettings {
|
|
@@ -71,9 +75,34 @@ export function createLMNT(options: LMNTProviderSettings = {}): LMNTProvider {
|
|
|
71
75
|
};
|
|
72
76
|
};
|
|
73
77
|
|
|
78
|
+
provider.specificationVersion = 'v4' as const;
|
|
74
79
|
provider.speech = createSpeechModel;
|
|
75
80
|
provider.speechModel = createSpeechModel;
|
|
76
81
|
|
|
82
|
+
provider.languageModel = (modelId: string) => {
|
|
83
|
+
throw new NoSuchModelError({
|
|
84
|
+
modelId,
|
|
85
|
+
modelType: 'languageModel',
|
|
86
|
+
message: 'LMNT does not provide language models',
|
|
87
|
+
});
|
|
88
|
+
};
|
|
89
|
+
|
|
90
|
+
provider.embeddingModel = (modelId: string) => {
|
|
91
|
+
throw new NoSuchModelError({
|
|
92
|
+
modelId,
|
|
93
|
+
modelType: 'embeddingModel',
|
|
94
|
+
message: 'LMNT does not provide embedding models',
|
|
95
|
+
});
|
|
96
|
+
};
|
|
97
|
+
|
|
98
|
+
provider.imageModel = (modelId: string) => {
|
|
99
|
+
throw new NoSuchModelError({
|
|
100
|
+
modelId,
|
|
101
|
+
modelType: 'imageModel',
|
|
102
|
+
message: 'LMNT does not provide image models',
|
|
103
|
+
});
|
|
104
|
+
};
|
|
105
|
+
|
|
77
106
|
return provider as LMNTProvider;
|
|
78
107
|
}
|
|
79
108
|
|
|
@@ -0,0 +1,69 @@
|
|
|
1
|
+
import { z } from 'zod/v4';
|
|
2
|
+
|
|
3
|
+
// https://docs.lmnt.com/api-reference/speech/synthesize-speech-bytes
|
|
4
|
+
export const lmntSpeechModelOptionsSchema = z.object({
|
|
5
|
+
/**
|
|
6
|
+
* The model to use for speech synthesis e.g. 'aurora' or 'blizzard'.
|
|
7
|
+
* @default 'aurora'
|
|
8
|
+
*/
|
|
9
|
+
model: z
|
|
10
|
+
.union([z.enum(['aurora', 'blizzard']), z.string()])
|
|
11
|
+
.nullish()
|
|
12
|
+
.default('aurora'),
|
|
13
|
+
|
|
14
|
+
/**
|
|
15
|
+
* The audio format of the output.
|
|
16
|
+
* @default 'mp3'
|
|
17
|
+
*/
|
|
18
|
+
format: z
|
|
19
|
+
.enum(['aac', 'mp3', 'mulaw', 'raw', 'wav'])
|
|
20
|
+
.nullish()
|
|
21
|
+
.default('mp3'),
|
|
22
|
+
|
|
23
|
+
/**
|
|
24
|
+
* The sample rate of the output audio in Hz.
|
|
25
|
+
* @default 24000
|
|
26
|
+
*/
|
|
27
|
+
sampleRate: z
|
|
28
|
+
.union([z.literal(8000), z.literal(16000), z.literal(24000)])
|
|
29
|
+
.nullish()
|
|
30
|
+
.default(24000),
|
|
31
|
+
|
|
32
|
+
/**
|
|
33
|
+
* The speed of the speech. Range: 0.25 to 2.
|
|
34
|
+
* @default 1
|
|
35
|
+
*/
|
|
36
|
+
speed: z.number().min(0.25).max(2).nullish().default(1),
|
|
37
|
+
|
|
38
|
+
/**
|
|
39
|
+
* A seed value for deterministic generation.
|
|
40
|
+
*/
|
|
41
|
+
seed: z.number().int().nullish(),
|
|
42
|
+
|
|
43
|
+
/**
|
|
44
|
+
* Whether to use a conversational style.
|
|
45
|
+
* @default false
|
|
46
|
+
*/
|
|
47
|
+
conversational: z.boolean().nullish().default(false),
|
|
48
|
+
|
|
49
|
+
/**
|
|
50
|
+
* Maximum length of the output in seconds (up to 300).
|
|
51
|
+
*/
|
|
52
|
+
length: z.number().max(300).nullish(),
|
|
53
|
+
|
|
54
|
+
/**
|
|
55
|
+
* Top-p sampling parameter. Range: 0 to 1.
|
|
56
|
+
* @default 1
|
|
57
|
+
*/
|
|
58
|
+
topP: z.number().min(0).max(1).nullish().default(1),
|
|
59
|
+
|
|
60
|
+
/**
|
|
61
|
+
* Temperature for sampling. Higher values increase randomness.
|
|
62
|
+
* @default 1
|
|
63
|
+
*/
|
|
64
|
+
temperature: z.number().min(0).nullish().default(1),
|
|
65
|
+
});
|
|
66
|
+
|
|
67
|
+
export type LMNTSpeechModelOptions = z.infer<
|
|
68
|
+
typeof lmntSpeechModelOptionsSchema
|
|
69
|
+
>;
|
package/src/lmnt-speech-model.ts
CHANGED
|
@@ -1,83 +1,18 @@
|
|
|
1
|
-
import {
|
|
1
|
+
import type { SpeechModelV4, SharedV4Warning } from '@ai-sdk/provider';
|
|
2
2
|
import {
|
|
3
3
|
combineHeaders,
|
|
4
4
|
createBinaryResponseHandler,
|
|
5
5
|
parseProviderOptions,
|
|
6
6
|
postJsonToApi,
|
|
7
|
+
serializeModelOptions,
|
|
8
|
+
WORKFLOW_SERIALIZE,
|
|
9
|
+
WORKFLOW_DESERIALIZE,
|
|
7
10
|
} from '@ai-sdk/provider-utils';
|
|
8
|
-
import {
|
|
9
|
-
import { LMNTConfig } from './lmnt-config';
|
|
11
|
+
import type { LMNTConfig } from './lmnt-config';
|
|
10
12
|
import { lmntFailedResponseHandler } from './lmnt-error';
|
|
11
|
-
import {
|
|
12
|
-
import {
|
|
13
|
-
|
|
14
|
-
// https://docs.lmnt.com/api-reference/speech/synthesize-speech-bytes
|
|
15
|
-
const lmntSpeechModelOptionsSchema = z.object({
|
|
16
|
-
/**
|
|
17
|
-
* The model to use for speech synthesis e.g. 'aurora' or 'blizzard'.
|
|
18
|
-
* @default 'aurora'
|
|
19
|
-
*/
|
|
20
|
-
model: z
|
|
21
|
-
.union([z.enum(['aurora', 'blizzard']), z.string()])
|
|
22
|
-
.nullish()
|
|
23
|
-
.default('aurora'),
|
|
24
|
-
|
|
25
|
-
/**
|
|
26
|
-
* The audio format of the output.
|
|
27
|
-
* @default 'mp3'
|
|
28
|
-
*/
|
|
29
|
-
format: z
|
|
30
|
-
.enum(['aac', 'mp3', 'mulaw', 'raw', 'wav'])
|
|
31
|
-
.nullish()
|
|
32
|
-
.default('mp3'),
|
|
33
|
-
|
|
34
|
-
/**
|
|
35
|
-
* The sample rate of the output audio in Hz.
|
|
36
|
-
* @default 24000
|
|
37
|
-
*/
|
|
38
|
-
sampleRate: z
|
|
39
|
-
.union([z.literal(8000), z.literal(16000), z.literal(24000)])
|
|
40
|
-
.nullish()
|
|
41
|
-
.default(24000),
|
|
42
|
-
|
|
43
|
-
/**
|
|
44
|
-
* The speed of the speech. Range: 0.25 to 2.
|
|
45
|
-
* @default 1
|
|
46
|
-
*/
|
|
47
|
-
speed: z.number().min(0.25).max(2).nullish().default(1),
|
|
48
|
-
|
|
49
|
-
/**
|
|
50
|
-
* A seed value for deterministic generation.
|
|
51
|
-
*/
|
|
52
|
-
seed: z.number().int().nullish(),
|
|
53
|
-
|
|
54
|
-
/**
|
|
55
|
-
* Whether to use a conversational style.
|
|
56
|
-
* @default false
|
|
57
|
-
*/
|
|
58
|
-
conversational: z.boolean().nullish().default(false),
|
|
59
|
-
|
|
60
|
-
/**
|
|
61
|
-
* Maximum length of the output in seconds (up to 300).
|
|
62
|
-
*/
|
|
63
|
-
length: z.number().max(300).nullish(),
|
|
64
|
-
|
|
65
|
-
/**
|
|
66
|
-
* Top-p sampling parameter. Range: 0 to 1.
|
|
67
|
-
* @default 1
|
|
68
|
-
*/
|
|
69
|
-
topP: z.number().min(0).max(1).nullish().default(1),
|
|
70
|
-
|
|
71
|
-
/**
|
|
72
|
-
* Temperature for sampling. Higher values increase randomness.
|
|
73
|
-
* @default 1
|
|
74
|
-
*/
|
|
75
|
-
temperature: z.number().min(0).nullish().default(1),
|
|
76
|
-
});
|
|
77
|
-
|
|
78
|
-
export type LMNTSpeechModelOptions = z.infer<
|
|
79
|
-
typeof lmntSpeechModelOptionsSchema
|
|
80
|
-
>;
|
|
13
|
+
import { lmntSpeechModelOptionsSchema } from './lmnt-speech-model-options';
|
|
14
|
+
import type { LMNTSpeechModelId } from './lmnt-speech-options';
|
|
15
|
+
import type { LMNTSpeechAPITypes } from './lmnt-api-types';
|
|
81
16
|
|
|
82
17
|
interface LMNTSpeechModelConfig extends LMNTConfig {
|
|
83
18
|
_internal?: {
|
|
@@ -85,13 +20,27 @@ interface LMNTSpeechModelConfig extends LMNTConfig {
|
|
|
85
20
|
};
|
|
86
21
|
}
|
|
87
22
|
|
|
88
|
-
export class LMNTSpeechModel implements
|
|
89
|
-
readonly specificationVersion = '
|
|
23
|
+
export class LMNTSpeechModel implements SpeechModelV4 {
|
|
24
|
+
readonly specificationVersion = 'v4';
|
|
90
25
|
|
|
91
26
|
get provider(): string {
|
|
92
27
|
return this.config.provider;
|
|
93
28
|
}
|
|
94
29
|
|
|
30
|
+
static [WORKFLOW_SERIALIZE](model: LMNTSpeechModel) {
|
|
31
|
+
return serializeModelOptions({
|
|
32
|
+
modelId: model.modelId,
|
|
33
|
+
config: model.config,
|
|
34
|
+
});
|
|
35
|
+
}
|
|
36
|
+
|
|
37
|
+
static [WORKFLOW_DESERIALIZE](options: {
|
|
38
|
+
modelId: LMNTSpeechModelId;
|
|
39
|
+
config: LMNTSpeechModelConfig;
|
|
40
|
+
}) {
|
|
41
|
+
return new LMNTSpeechModel(options.modelId, options.config);
|
|
42
|
+
}
|
|
43
|
+
|
|
95
44
|
constructor(
|
|
96
45
|
readonly modelId: LMNTSpeechModelId,
|
|
97
46
|
private readonly config: LMNTSpeechModelConfig,
|
|
@@ -104,8 +53,8 @@ export class LMNTSpeechModel implements SpeechModelV3 {
|
|
|
104
53
|
speed,
|
|
105
54
|
language,
|
|
106
55
|
providerOptions,
|
|
107
|
-
}: Parameters<
|
|
108
|
-
const warnings:
|
|
56
|
+
}: Parameters<SpeechModelV4['doGenerate']>[0]) {
|
|
57
|
+
const warnings: SharedV4Warning[] = [];
|
|
109
58
|
|
|
110
59
|
// Parse provider options
|
|
111
60
|
const lmntOptions = await parseProviderOptions({
|
|
@@ -169,8 +118,8 @@ export class LMNTSpeechModel implements SpeechModelV3 {
|
|
|
169
118
|
}
|
|
170
119
|
|
|
171
120
|
async doGenerate(
|
|
172
|
-
options: Parameters<
|
|
173
|
-
): Promise<Awaited<ReturnType<
|
|
121
|
+
options: Parameters<SpeechModelV4['doGenerate']>[0],
|
|
122
|
+
): Promise<Awaited<ReturnType<SpeechModelV4['doGenerate']>>> {
|
|
174
123
|
const currentDate = this.config._internal?.currentDate?.() ?? new Date();
|
|
175
124
|
const { requestBody, warnings } = await this.getArgs(options);
|
|
176
125
|
|
|
@@ -183,7 +132,7 @@ export class LMNTSpeechModel implements SpeechModelV3 {
|
|
|
183
132
|
path: '/v1/ai/speech/bytes',
|
|
184
133
|
modelId: this.modelId,
|
|
185
134
|
}),
|
|
186
|
-
headers: combineHeaders(this.config.headers(), options.headers),
|
|
135
|
+
headers: combineHeaders(this.config.headers?.(), options.headers),
|
|
187
136
|
body: requestBody,
|
|
188
137
|
failedResponseHandler: lmntFailedResponseHandler,
|
|
189
138
|
successfulResponseHandler: createBinaryResponseHandler(),
|
package/dist/index.d.mts
DELETED
|
@@ -1,89 +0,0 @@
|
|
|
1
|
-
import { SpeechModelV3, ProviderV3 } from '@ai-sdk/provider';
|
|
2
|
-
import { FetchFunction } from '@ai-sdk/provider-utils';
|
|
3
|
-
import { z } from 'zod/v4';
|
|
4
|
-
|
|
5
|
-
type LMNTConfig = {
|
|
6
|
-
provider: string;
|
|
7
|
-
url: (options: {
|
|
8
|
-
modelId: string;
|
|
9
|
-
path: string;
|
|
10
|
-
}) => string;
|
|
11
|
-
headers: () => Record<string, string | undefined>;
|
|
12
|
-
fetch?: FetchFunction;
|
|
13
|
-
generateId?: () => string;
|
|
14
|
-
};
|
|
15
|
-
|
|
16
|
-
type LMNTSpeechModelId = 'aurora' | 'blizzard' | (string & {});
|
|
17
|
-
|
|
18
|
-
declare const lmntSpeechModelOptionsSchema: z.ZodObject<{
|
|
19
|
-
model: z.ZodDefault<z.ZodOptional<z.ZodNullable<z.ZodUnion<readonly [z.ZodEnum<{
|
|
20
|
-
aurora: "aurora";
|
|
21
|
-
blizzard: "blizzard";
|
|
22
|
-
}>, z.ZodString]>>>>;
|
|
23
|
-
format: z.ZodDefault<z.ZodOptional<z.ZodNullable<z.ZodEnum<{
|
|
24
|
-
aac: "aac";
|
|
25
|
-
mp3: "mp3";
|
|
26
|
-
mulaw: "mulaw";
|
|
27
|
-
raw: "raw";
|
|
28
|
-
wav: "wav";
|
|
29
|
-
}>>>>;
|
|
30
|
-
sampleRate: z.ZodDefault<z.ZodOptional<z.ZodNullable<z.ZodUnion<readonly [z.ZodLiteral<8000>, z.ZodLiteral<16000>, z.ZodLiteral<24000>]>>>>;
|
|
31
|
-
speed: z.ZodDefault<z.ZodOptional<z.ZodNullable<z.ZodNumber>>>;
|
|
32
|
-
seed: z.ZodOptional<z.ZodNullable<z.ZodNumber>>;
|
|
33
|
-
conversational: z.ZodDefault<z.ZodOptional<z.ZodNullable<z.ZodBoolean>>>;
|
|
34
|
-
length: z.ZodOptional<z.ZodNullable<z.ZodNumber>>;
|
|
35
|
-
topP: z.ZodDefault<z.ZodOptional<z.ZodNullable<z.ZodNumber>>>;
|
|
36
|
-
temperature: z.ZodDefault<z.ZodOptional<z.ZodNullable<z.ZodNumber>>>;
|
|
37
|
-
}, z.core.$strip>;
|
|
38
|
-
type LMNTSpeechModelOptions = z.infer<typeof lmntSpeechModelOptionsSchema>;
|
|
39
|
-
interface LMNTSpeechModelConfig extends LMNTConfig {
|
|
40
|
-
_internal?: {
|
|
41
|
-
currentDate?: () => Date;
|
|
42
|
-
};
|
|
43
|
-
}
|
|
44
|
-
declare class LMNTSpeechModel implements SpeechModelV3 {
|
|
45
|
-
readonly modelId: LMNTSpeechModelId;
|
|
46
|
-
private readonly config;
|
|
47
|
-
readonly specificationVersion = "v3";
|
|
48
|
-
get provider(): string;
|
|
49
|
-
constructor(modelId: LMNTSpeechModelId, config: LMNTSpeechModelConfig);
|
|
50
|
-
private getArgs;
|
|
51
|
-
doGenerate(options: Parameters<SpeechModelV3['doGenerate']>[0]): Promise<Awaited<ReturnType<SpeechModelV3['doGenerate']>>>;
|
|
52
|
-
}
|
|
53
|
-
|
|
54
|
-
interface LMNTProvider extends Pick<ProviderV3, 'speechModel'> {
|
|
55
|
-
(modelId: 'aurora', settings?: {}): {
|
|
56
|
-
speech: LMNTSpeechModel;
|
|
57
|
-
};
|
|
58
|
-
/**
|
|
59
|
-
* Creates a model for speech synthesis.
|
|
60
|
-
*/
|
|
61
|
-
speech(modelId: LMNTSpeechModelId): SpeechModelV3;
|
|
62
|
-
}
|
|
63
|
-
interface LMNTProviderSettings {
|
|
64
|
-
/**
|
|
65
|
-
* API key for authenticating requests.
|
|
66
|
-
*/
|
|
67
|
-
apiKey?: string;
|
|
68
|
-
/**
|
|
69
|
-
* Custom headers to include in the requests.
|
|
70
|
-
*/
|
|
71
|
-
headers?: Record<string, string>;
|
|
72
|
-
/**
|
|
73
|
-
* Custom fetch implementation. You can use it as a middleware to intercept requests,
|
|
74
|
-
* or to provide a custom fetch implementation for e.g. testing.
|
|
75
|
-
*/
|
|
76
|
-
fetch?: FetchFunction;
|
|
77
|
-
}
|
|
78
|
-
/**
|
|
79
|
-
* Create an LMNT provider instance.
|
|
80
|
-
*/
|
|
81
|
-
declare function createLMNT(options?: LMNTProviderSettings): LMNTProvider;
|
|
82
|
-
/**
|
|
83
|
-
* Default LMNT provider instance.
|
|
84
|
-
*/
|
|
85
|
-
declare const lmnt: LMNTProvider;
|
|
86
|
-
|
|
87
|
-
declare const VERSION: string;
|
|
88
|
-
|
|
89
|
-
export { type LMNTProvider, type LMNTProviderSettings, type LMNTSpeechModelOptions, VERSION, createLMNT, lmnt };
|
package/dist/index.mjs
DELETED
|
@@ -1,216 +0,0 @@
|
|
|
1
|
-
// src/lmnt-provider.ts
|
|
2
|
-
import {
|
|
3
|
-
loadApiKey,
|
|
4
|
-
withUserAgentSuffix
|
|
5
|
-
} from "@ai-sdk/provider-utils";
|
|
6
|
-
|
|
7
|
-
// src/lmnt-speech-model.ts
|
|
8
|
-
import {
|
|
9
|
-
combineHeaders,
|
|
10
|
-
createBinaryResponseHandler,
|
|
11
|
-
parseProviderOptions,
|
|
12
|
-
postJsonToApi
|
|
13
|
-
} from "@ai-sdk/provider-utils";
|
|
14
|
-
import { z as z2 } from "zod/v4";
|
|
15
|
-
|
|
16
|
-
// src/lmnt-error.ts
|
|
17
|
-
import { z } from "zod/v4";
|
|
18
|
-
import { createJsonErrorResponseHandler } from "@ai-sdk/provider-utils";
|
|
19
|
-
var lmntErrorDataSchema = z.object({
|
|
20
|
-
error: z.object({
|
|
21
|
-
message: z.string(),
|
|
22
|
-
code: z.number()
|
|
23
|
-
})
|
|
24
|
-
});
|
|
25
|
-
var lmntFailedResponseHandler = createJsonErrorResponseHandler({
|
|
26
|
-
errorSchema: lmntErrorDataSchema,
|
|
27
|
-
errorToMessage: (data) => data.error.message
|
|
28
|
-
});
|
|
29
|
-
|
|
30
|
-
// src/lmnt-speech-model.ts
|
|
31
|
-
var lmntSpeechModelOptionsSchema = z2.object({
|
|
32
|
-
/**
|
|
33
|
-
* The model to use for speech synthesis e.g. 'aurora' or 'blizzard'.
|
|
34
|
-
* @default 'aurora'
|
|
35
|
-
*/
|
|
36
|
-
model: z2.union([z2.enum(["aurora", "blizzard"]), z2.string()]).nullish().default("aurora"),
|
|
37
|
-
/**
|
|
38
|
-
* The audio format of the output.
|
|
39
|
-
* @default 'mp3'
|
|
40
|
-
*/
|
|
41
|
-
format: z2.enum(["aac", "mp3", "mulaw", "raw", "wav"]).nullish().default("mp3"),
|
|
42
|
-
/**
|
|
43
|
-
* The sample rate of the output audio in Hz.
|
|
44
|
-
* @default 24000
|
|
45
|
-
*/
|
|
46
|
-
sampleRate: z2.union([z2.literal(8e3), z2.literal(16e3), z2.literal(24e3)]).nullish().default(24e3),
|
|
47
|
-
/**
|
|
48
|
-
* The speed of the speech. Range: 0.25 to 2.
|
|
49
|
-
* @default 1
|
|
50
|
-
*/
|
|
51
|
-
speed: z2.number().min(0.25).max(2).nullish().default(1),
|
|
52
|
-
/**
|
|
53
|
-
* A seed value for deterministic generation.
|
|
54
|
-
*/
|
|
55
|
-
seed: z2.number().int().nullish(),
|
|
56
|
-
/**
|
|
57
|
-
* Whether to use a conversational style.
|
|
58
|
-
* @default false
|
|
59
|
-
*/
|
|
60
|
-
conversational: z2.boolean().nullish().default(false),
|
|
61
|
-
/**
|
|
62
|
-
* Maximum length of the output in seconds (up to 300).
|
|
63
|
-
*/
|
|
64
|
-
length: z2.number().max(300).nullish(),
|
|
65
|
-
/**
|
|
66
|
-
* Top-p sampling parameter. Range: 0 to 1.
|
|
67
|
-
* @default 1
|
|
68
|
-
*/
|
|
69
|
-
topP: z2.number().min(0).max(1).nullish().default(1),
|
|
70
|
-
/**
|
|
71
|
-
* Temperature for sampling. Higher values increase randomness.
|
|
72
|
-
* @default 1
|
|
73
|
-
*/
|
|
74
|
-
temperature: z2.number().min(0).nullish().default(1)
|
|
75
|
-
});
|
|
76
|
-
var LMNTSpeechModel = class {
|
|
77
|
-
constructor(modelId, config) {
|
|
78
|
-
this.modelId = modelId;
|
|
79
|
-
this.config = config;
|
|
80
|
-
this.specificationVersion = "v3";
|
|
81
|
-
}
|
|
82
|
-
get provider() {
|
|
83
|
-
return this.config.provider;
|
|
84
|
-
}
|
|
85
|
-
async getArgs({
|
|
86
|
-
text,
|
|
87
|
-
voice = "ava",
|
|
88
|
-
outputFormat = "mp3",
|
|
89
|
-
speed,
|
|
90
|
-
language,
|
|
91
|
-
providerOptions
|
|
92
|
-
}) {
|
|
93
|
-
var _a, _b, _c, _d, _e, _f, _g;
|
|
94
|
-
const warnings = [];
|
|
95
|
-
const lmntOptions = await parseProviderOptions({
|
|
96
|
-
provider: "lmnt",
|
|
97
|
-
providerOptions,
|
|
98
|
-
schema: lmntSpeechModelOptionsSchema
|
|
99
|
-
});
|
|
100
|
-
const requestBody = {
|
|
101
|
-
model: this.modelId,
|
|
102
|
-
text,
|
|
103
|
-
voice,
|
|
104
|
-
response_format: "mp3",
|
|
105
|
-
speed
|
|
106
|
-
};
|
|
107
|
-
if (outputFormat) {
|
|
108
|
-
if (["mp3", "aac", "mulaw", "raw", "wav"].includes(outputFormat)) {
|
|
109
|
-
requestBody.response_format = outputFormat;
|
|
110
|
-
} else {
|
|
111
|
-
warnings.push({
|
|
112
|
-
type: "unsupported",
|
|
113
|
-
feature: "outputFormat",
|
|
114
|
-
details: `Unsupported output format: ${outputFormat}. Using mp3 instead.`
|
|
115
|
-
});
|
|
116
|
-
}
|
|
117
|
-
}
|
|
118
|
-
if (lmntOptions) {
|
|
119
|
-
const speechModelOptions = {
|
|
120
|
-
conversational: (_a = lmntOptions.conversational) != null ? _a : void 0,
|
|
121
|
-
length: (_b = lmntOptions.length) != null ? _b : void 0,
|
|
122
|
-
seed: (_c = lmntOptions.seed) != null ? _c : void 0,
|
|
123
|
-
speed: (_d = lmntOptions.speed) != null ? _d : void 0,
|
|
124
|
-
temperature: (_e = lmntOptions.temperature) != null ? _e : void 0,
|
|
125
|
-
top_p: (_f = lmntOptions.topP) != null ? _f : void 0,
|
|
126
|
-
sample_rate: (_g = lmntOptions.sampleRate) != null ? _g : void 0
|
|
127
|
-
};
|
|
128
|
-
for (const key in speechModelOptions) {
|
|
129
|
-
const value = speechModelOptions[key];
|
|
130
|
-
if (value !== void 0) {
|
|
131
|
-
requestBody[key] = value;
|
|
132
|
-
}
|
|
133
|
-
}
|
|
134
|
-
}
|
|
135
|
-
if (language) {
|
|
136
|
-
requestBody.language = language;
|
|
137
|
-
}
|
|
138
|
-
return {
|
|
139
|
-
requestBody,
|
|
140
|
-
warnings
|
|
141
|
-
};
|
|
142
|
-
}
|
|
143
|
-
async doGenerate(options) {
|
|
144
|
-
var _a, _b, _c;
|
|
145
|
-
const currentDate = (_c = (_b = (_a = this.config._internal) == null ? void 0 : _a.currentDate) == null ? void 0 : _b.call(_a)) != null ? _c : /* @__PURE__ */ new Date();
|
|
146
|
-
const { requestBody, warnings } = await this.getArgs(options);
|
|
147
|
-
const {
|
|
148
|
-
value: audio,
|
|
149
|
-
responseHeaders,
|
|
150
|
-
rawValue: rawResponse
|
|
151
|
-
} = await postJsonToApi({
|
|
152
|
-
url: this.config.url({
|
|
153
|
-
path: "/v1/ai/speech/bytes",
|
|
154
|
-
modelId: this.modelId
|
|
155
|
-
}),
|
|
156
|
-
headers: combineHeaders(this.config.headers(), options.headers),
|
|
157
|
-
body: requestBody,
|
|
158
|
-
failedResponseHandler: lmntFailedResponseHandler,
|
|
159
|
-
successfulResponseHandler: createBinaryResponseHandler(),
|
|
160
|
-
abortSignal: options.abortSignal,
|
|
161
|
-
fetch: this.config.fetch
|
|
162
|
-
});
|
|
163
|
-
return {
|
|
164
|
-
audio,
|
|
165
|
-
warnings,
|
|
166
|
-
request: {
|
|
167
|
-
body: JSON.stringify(requestBody)
|
|
168
|
-
},
|
|
169
|
-
response: {
|
|
170
|
-
timestamp: currentDate,
|
|
171
|
-
modelId: this.modelId,
|
|
172
|
-
headers: responseHeaders,
|
|
173
|
-
body: rawResponse
|
|
174
|
-
}
|
|
175
|
-
};
|
|
176
|
-
}
|
|
177
|
-
};
|
|
178
|
-
|
|
179
|
-
// src/version.ts
|
|
180
|
-
var VERSION = true ? "3.0.0-beta.3" : "0.0.0-test";
|
|
181
|
-
|
|
182
|
-
// src/lmnt-provider.ts
|
|
183
|
-
function createLMNT(options = {}) {
|
|
184
|
-
const getHeaders = () => withUserAgentSuffix(
|
|
185
|
-
{
|
|
186
|
-
"x-api-key": loadApiKey({
|
|
187
|
-
apiKey: options.apiKey,
|
|
188
|
-
environmentVariableName: "LMNT_API_KEY",
|
|
189
|
-
description: "LMNT"
|
|
190
|
-
}),
|
|
191
|
-
...options.headers
|
|
192
|
-
},
|
|
193
|
-
`ai-sdk/lmnt/${VERSION}`
|
|
194
|
-
);
|
|
195
|
-
const createSpeechModel = (modelId) => new LMNTSpeechModel(modelId, {
|
|
196
|
-
provider: `lmnt.speech`,
|
|
197
|
-
url: ({ path }) => `https://api.lmnt.com${path}`,
|
|
198
|
-
headers: getHeaders,
|
|
199
|
-
fetch: options.fetch
|
|
200
|
-
});
|
|
201
|
-
const provider = function(modelId) {
|
|
202
|
-
return {
|
|
203
|
-
speech: createSpeechModel(modelId)
|
|
204
|
-
};
|
|
205
|
-
};
|
|
206
|
-
provider.speech = createSpeechModel;
|
|
207
|
-
provider.speechModel = createSpeechModel;
|
|
208
|
-
return provider;
|
|
209
|
-
}
|
|
210
|
-
var lmnt = createLMNT();
|
|
211
|
-
export {
|
|
212
|
-
VERSION,
|
|
213
|
-
createLMNT,
|
|
214
|
-
lmnt
|
|
215
|
-
};
|
|
216
|
-
//# sourceMappingURL=index.mjs.map
|