sarvam-ai-sdk 0.1.2 → 0.1.4

package/README.md

@@ -1,219 +1,245 @@
-# AI SDK - Sarvam Provider
-
-The **[Sarvam provider](https://v4.ai-sdk.dev/providers/ai-sdk-providers/sarvam)** for the [AI SDK](https://v4.ai-sdk.dev/docs)
-contains language model support for the Sarvam chat completion, Text-to-Speech and Speech-to-Text APIs.
-
-## Setup
-
-The **[Sarvam](http://sarvam.ai)** provider is available in the `sarvam-ai-sdk` module. You can install it with
-
-```bash
-npm i sarvam-ai-sdk
-```
-
-> [!WARNING]
-> This package only works with Vercel AI-SDK v4, not latest v6. Make sure to install `ai@4` in your project.
-
-## Provider Instance
-
-You can import the default provider instance `sarvam` from `sarvam-ai-sdk`:
-
-```ts
-import { sarvam } from 'sarvam-ai-sdk';
-```
-
-Create `.env` file with API key from **[Sarvam Dashboard](https://dashboard.sarvam.ai/)**
-```bash
-SARVAM_API_KEY="your_api_key"
-```
-
-## Example
-
-```ts
-import { sarvam } from 'sarvam-ai-sdk';
-import { generateText } from 'ai';
-
-const { text } = await generateText({
-	model: sarvam("sarvam-30b"),
-    prompt: "Translate this to malayalam: 'Keep cooking, guys'",
-});
-
-console.log(text); // പാചകം തുടരൂ, സുഹൃത്തുക്കളേ
-```
-
-## Text-to-Speech
-
-```ts
-import { sarvam } from "sarvam-ai-sdk";
-import { experimental_generateSpeech as generateSpeech } from "ai";
-import { writeFile } from "fs/promises";
-
-const { audio } = await generateSpeech({
-    model: sarvam.speech("bulbul:v3", "ml-IN"),
-    text: "പാചകം തുടരൂ, സുഹൃത്തുക്കളേ",
-});
-
-const audioBuffer = Buffer.from(audio.base64, "base64")
-await writeFile("./src/transcript-test.wav", audioBuffer);
-```
-
-## Speech-to-Text
-
-```ts
-import { sarvam } from "sarvam-ai-sdk";
-import { experimental_transcribe as transcribe } from "ai";
-import { readFile } from "fs/promises";
-
-const { text } = await transcribe({
-    model: sarvam.transcription("saarika:v2.5", "ml-IN")
-    audio: await readFile("./src/transcript-test.wav"),
-});
-
-console.log(text); // പാചകം തുടരും സുഹൃത്തുക്കളെ
-```
-
-```ts
-import { sarvam } from "sarvam-ai-sdk";
-import { experimental_transcribe as transcribe } from "ai";
-import { readFile } from "fs/promises";
-
-const { text } = await transcribe({
-    model: sarvam.transcription("saaras:v3", "en-IN"),
-    audio: await readFile("./src/transcript-test.wav"),
-});
-
-console.log(text); // Pachakam thudaroo, suhruthukkale.
-```
-
-## Speech-to-Text-Translate
-
-```ts
-import { sarvam } from "sarvam-ai-sdk";
-import { experimental_transcribe as transcribe } from "ai";
-import { readFile } from "fs/promises";
-
-const result = await transcribe({
-    model: sarvam.speechTranslation("saaras:v2.5"),
-    audio: await readFile("./src/transcript-test.wav"),
-});
-
-console.log(result.text); // Cooking continues, my friends
-```
-
-## Translation
-
-> NB: Only transliterates `prompt` and `role:user` messages, not `system` not `assistant`.
-
-```ts
-import { sarvam } from "sarvam-ai-sdk";
-import { generateText } from "ai";
-
-const result = await generateText({
-    model: sarvam.translation({
-        "from": "ml-IN",
-        "to": "en-IN",
-    }),
-    prompt: "ഇതൊക്കെ ശ്രദ്ധിക്കണ്ടേ അംബാനെ?",
-});
-
-console.log(result.text); // Shouldn't we be careful about this, Ambane?
-```
-
-## Transliterate
-
-> NB: Only transliterates `prompt` and `role:user` messages, not `system` not `assistant`.
-
-```ts
-import { sarvam } from "sarvam-ai-sdk";
-import { generateText } from "ai";
-
-const result = await generateText({
-  model: sarvam.transliterate({
-      from: "en-IN",
-      to: "ml-IN",
-  }),
-  prompt: "eda mone, happy alle?",
-});
-
-console.log(result.text); // എടാ മോനെ, ഹാപ്പി അല്ലേ?
-```
-
-## Language Identification
-
-> NB: Only identifies `prompt` and `role:user` messages, not `system` not `assistant`.
-
-```ts
-import { sarvam } from "sarvam-ai-sdk";
-import { generateText } from "ai";
-
-const result = await generateText({
-    model: sarvam.languageIdentification(),
-    prompt: "ബുദ്ധിയാണ് സാറേ ഇവൻ്റെ മെയിൻ",
-});
-
-console.log(result.text); // ml-IN
-```
-
-## Tool Calling
-
-> [!WARNING]
-> Latest `sarvam` models isn't trained on native tool calling feature (aka JSON mode). So we simulate this with prompt engineering technique.
-
-```ts
-import { z } from "zod";
-import { generateText, tool } from "ai";
-import { sarvam } from "sarvam-ai-sdk";
-
-
-const result = await generateText({
-  model: sarvam("sarvam-30b", {
-    simulate: "tool-calling" // ⚠️ important
-  }),
-  tools: {
-    weather: tool({
-      description: "Get the weather in a location",
-      parameters: z.object({
-        location: z.string().describe("The location to get the weather for"),
-      }),
-      execute: async ({ location }) => ({
-        location,
-        temperature: 72 + Math.floor(Math.random() * 21) - 10,
-      }),
-    }),
-  },
-  system: "Your are a helpful AI",
-  prompt: "കൊച്ചിയിലെ കാലാവസ്ഥ എന്താണ്?",
-});
-
-console.log(result.toolResults);
-```
-## Generate JSON object
-
-> [!WARNING]
-> Latest `sarvam` models isn't trained on native JSON object generation. So we simulate this with prompt engineering technique.
-
-```ts
-import { z } from "zod";
-import { sarvam } from "sarvam-ai-sdk";
-import { generateObject } from 'ai';
-
-const { object } = await generateObject({
-  model: sarvam("sarvam-30b", {
-    simulate: "json-object" // ⚠️ important
-  }),
-  schema: z.object({
-    recipe: z.object({
-      name: z.string(),
-      ingredients: z.array(z.string()),
-      steps: z.array(z.string()),
-    }),
-  }),
-  prompt: 'Generate a South Indian recipe, in Malayalam',
-});
-
-console.log(object);
-```
-
-## Documentation
-
-Please check out the **[Sarvam provider documentation](https://v4.ai-sdk.dev/providers/ai-sdk-providers/sarvam)** and **[Sarvam API documentation](https://docs.sarvam.ai)** for more information.
+# AI SDK - Sarvam Provider
+
+The **[Sarvam provider](https://v4.ai-sdk.dev/providers/ai-sdk-providers/sarvam)** for the [AI SDK](https://v4.ai-sdk.dev/docs)
+contains language model support for the Sarvam chat completion, Text-to-Speech and Speech-to-Text APIs.
+
+## Setup
+
+The **[Sarvam](http://sarvam.ai)** provider is available in the `sarvam-ai-sdk` module. You can install it with
+
+```bash
+npm i sarvam-ai-sdk
+```
+
+> [!WARNING]
+> This package only works with Vercel AI-SDK v4, not latest v6. Make sure to install `ai@4` in your project.
+
+## Provider Instance
+
+You can import the default provider instance `sarvam` from `sarvam-ai-sdk`:
+
+```ts
+import { sarvam } from 'sarvam-ai-sdk';
+```
+
+Create `.env` file with API key from **[Sarvam Dashboard](https://dashboard.sarvam.ai/)**
+```bash
+SARVAM_API_KEY="your_api_key"
+```
+
+## Example
+
+```ts
+import { sarvam } from 'sarvam-ai-sdk';
+import { generateText } from 'ai';
+
+const { text } = await generateText({
+	model: sarvam("sarvam-30b"),
+    prompt: "Translate this to malayalam: 'Keep cooking, guys'",
+});
+
+console.log(text); // പാചകം തുടരൂ, സുഹൃത്തുക്കളേ
+```
+
+## Text-to-Speech
+
+```ts
+import { sarvam } from "sarvam-ai-sdk";
+import { experimental_generateSpeech as generateSpeech } from "ai";
+import { writeFile } from "fs/promises";
+
+const { audio } = await generateSpeech({
+    model: sarvam.speech("bulbul:v3", "ml-IN"),
+    text: "പാചകം തുടരൂ, സുഹൃത്തുക്കളേ",
+});
+
+const audioBuffer = Buffer.from(audio.base64, "base64")
+await writeFile("./src/transcript-test.wav", audioBuffer);
+```
+
+## Speech-to-Text
+
+```ts
+import { sarvam } from "sarvam-ai-sdk";
+import { experimental_transcribe as transcribe } from "ai";
+import { readFile } from "fs/promises";
+
+const { text } = await transcribe({
+    model: sarvam.transcription("saarika:v2.5", "ml-IN"),
+    audio: await readFile("./src/transcript-test.wav"),
+});
+
+console.log(text); // പാചകം തുടരും സുഹൃത്തുക്കളെ
+```
+
+```ts
+import { sarvam } from "sarvam-ai-sdk";
+import { experimental_transcribe as transcribe } from "ai";
+import { readFile } from "fs/promises";
+
+const { text } = await transcribe({
+    model: sarvam.transcription("saaras:v3", "en-IN"),
+    audio: await readFile("./src/transcript-test.wav"),
+});
+
+console.log(text); // Pachakam thudaroo, suhruthukkale.
+```
+
+## Speech-to-Text-Translate
+
+```ts
+import { sarvam } from "sarvam-ai-sdk";
+import { experimental_transcribe as transcribe } from "ai";
+import { readFile } from "fs/promises";
+
+const result = await transcribe({
+    model: sarvam.speechTranslation("saaras:v2.5"),
+    audio: await readFile("./src/transcript-test.wav"),
+});
+
+console.log(result.text); // Cooking continues, my friends
+```
+
+## Translation
+
+> NB: Only transliterates `prompt` and `role:user` messages, not `system` not `assistant`.
+
+```ts
+import { sarvam } from "sarvam-ai-sdk";
+import { generateText } from "ai";
+
+const result = await generateText({
+    model: sarvam.translation({
+        "from": "ml-IN",
+        "to": "en-IN",
+    }),
+    prompt: "ഇതൊക്കെ ശ്രദ്ധിക്കണ്ടേ അംബാനെ?",
+});
+
+console.log(result.text); // Shouldn't we be careful about this, Ambane?
+```
+
+## Transliterate
+
+> NB: Only transliterates `prompt` and `role:user` messages, not `system` not `assistant`.
+
+```ts
+import { sarvam } from "sarvam-ai-sdk";
+import { generateText } from "ai";
+
+const result = await generateText({
+  model: sarvam.transliterate({
+      from: "en-IN",
+      to: "ml-IN",
+  }),
+  prompt: "eda mone, happy alle?",
+});
+
+console.log(result.text); // എടാ മോനെ, ഹാപ്പി അല്ലേ?
+```
+
+## Language Identification
+
+> NB: Only identifies `prompt` and `role:user` messages, not `system` not `assistant`.
+
+```ts
+import { sarvam } from "sarvam-ai-sdk";
+import { generateText } from "ai";
+
+const result = await generateText({
+    model: sarvam.languageIdentification(),
+    prompt: "ബുദ്ധിയാണ് സാറേ ഇവൻ്റെ മെയിൻ",
+});
+
+console.log(result.text); // ml-IN
+```
+
+## Tool Calling
+
+```ts
+import { z } from "zod";
+import { generateText, tool } from "ai";
+import { sarvam } from "sarvam-ai-sdk";
+
+
+const result = await generateText({
+  model: sarvam("sarvam-30b"),
+  tools: {
+    weather: tool({
+      description: "Get the weather in a location",
+      parameters: z.object({
+        location: z.string().describe("The location to get the weather for"),
+      }),
+      execute: async ({ location }) => ({
+        location,
+        temperature: 72 + Math.floor(Math.random() * 21) - 10,
+      }),
+    }),
+  },
+  system: "Your are a helpful AI",
+  prompt: "കൊച്ചിയിലെ കാലാവസ്ഥ എന്താണ്?",
+});
+
+console.log(result.toolResults);
+```
+
+> [!WARNING]
+> Old `sarvam-m` models isn't trained on native tool calling feature (aka JSON mode). So we recommend using latest models.
+
+## Generate JSON object
+
+```ts
+import { z } from "zod";
+import { sarvam } from "sarvam-ai-sdk";
+import { generateObject } from 'ai';
+
+const { object } = await generateObject({
+  model: sarvam("sarvam-30b"),
+  schema: z.object({
+    recipe: z.object({
+      name: z.string(),
+      ingredients: z.array(z.string()),
+      steps: z.array(z.string()),
+    }),
+  }),
+  prompt: 'Generate a South Indian recipe, in Malayalam',
+});
+
+console.log(object);
+```
+
+> [!WARNING]
+> Old `sarvam-m` models isn't trained on native JSON object generation. So we recommend using latest models.
+
+## All APIs
+
+```ts
+import { sarvam } from "sarvam-ai-sdk";
+
+// Text-to-Text + Chat Completion
+sarvam("sarvam-105b");
+sarvam.languageModel("sarvam-30b");
+
+// Text-to-Text + Transliteration
+sarvam.transliterate({ from: "en-IN", to: "ml-IN" });
+
+// Text-to-Text + Translation
+sarvam.translation({ from: "en-IN", to: "ml-IN" });
+
+// Text-to-Text + Language identification
+sarvam.languageIdentification();
+
+// Text-to-Speech
+sarvam.speech("bulbul:v3", "ml-IN");
+
+// Speech-to-Text + Transcribe to same language
+sarvam.transcription("saarika:v2.5");
+
+// Speech-to-Text + Translate to English
+sarvam.speechTranslation("saaras:v3");
+```
+
+
+## Documentation
+
+Please check out the **[Sarvam provider documentation](https://v4.ai-sdk.dev/providers/ai-sdk-providers/sarvam)** and **[Sarvam API documentation](https://docs.sarvam.ai)** for more information.

package/dist/index.d.mts

@@ -3,34 +3,10 @@ import { FetchFunction } from '@ai-sdk/provider-utils';
 import { z } from 'zod';
 
 /**
- * @description Product models
+ * @description Production models
  */
-type SarvamChatModelId = "sarvam-30b" | "sarvam-30b-16k" | "sarvam-105b" | "sarvam-105b-32k" | SarvamChatLegacyModelId | (string & {});
-/**
- * @description Legacy models
- * @deprecated
- */
-type SarvamChatLegacyModelId = "sarvam-m";
+type SarvamChatModelId = "sarvam-30b" | "sarvam-30b-16k" | "sarvam-105b" | "sarvam-105b-32k" | (string & {});
 interface SarvamChatSettings {
-    /**
-    * Whether to simulate artificial tool calling or JSON object generation, because Sarvam Models doen't support native Tool Calling or JSON Schmea.
-    * @default undefined
-    * @example
-        await generateText({
-            model: sarvam("sarvam-m", {
-                simulate: "tool-calling"
-            })
-            tools: {...}
-        })
-
-        await generateObject({
-            model: sarvam("sarvam-m", {
-                simulate: "json-object"
-            })
-            schema: {...}
-        })
-    */
-    simulate?: "tool-calling" | "json-object";
     /**
      * Whether to enable parallel function calling during tool use.
      * @default true
@@ -52,7 +28,7 @@ declare const SarvamLanguageCodeSchema: z.ZodEnum<["hi-IN", "bn-IN", "kn-IN", "m
 
 type SarvamSpeechModelId = "bulbul:v2" | "bulbul:v3" | (string & {});
 type SarvamSpeechVoices = z.infer<typeof SpeakerSchema>;
-declare const SpeakerSchema: z.ZodDefault<z.ZodEnum<["abhilash", "karun", "hitesh", "anushka", "manisha", "vidya", "arya", "shubh", "aditya", "rahul", "rohan", "amit", "dev", "ratan", "varun", "manan", "sumit", "kabir", "aayan", "ashutosh", "advait", "anand", "tarun", "sunny", "mani", "gokul", "vijay", "mohit", "rehan", "soham", "ritu", "priya", "neha", "pooja", "simran", "kavya", "ishita", "shreya", "roopa", "amelia", "sophia", "tanya", "shruti", "suhani", "kavitha", "rupali"]>>;
+declare const SpeakerSchema: z.ZodEnum<["abhilash", "karun", "hitesh", "anushka", "manisha", "vidya", "arya", "shubh", "aditya", "rahul", "rohan", "amit", "dev", "ratan", "varun", "manan", "sumit", "kabir", "aayan", "ashutosh", "advait", "anand", "tarun", "sunny", "mani", "gokul", "vijay", "mohit", "rehan", "soham", "ritu", "priya", "neha", "pooja", "simran", "kavya", "ishita", "shreya", "roopa", "amelia", "sophia", "tanya", "shruti", "suhani", "kavitha", "rupali"]>;
 /**
  * Configuration settings for Sarvam Text-to-Speech API.
  *
@@ -111,6 +87,37 @@ type SarvamSpeechSettings = {
      * @example false (Disable preprocessing)
      */
     enable_preprocessing?: boolean;
+    /**
+     * Specifies the audio codec for the output audio file.
+     * Different codecs offer various compression and quality characteristics.
+     */
+    output_audio_codec?: "mp3" | "linear16" | "mulaw" | "alaw" | "opus" | "flac" | "aac" | "wav";
+    /**
+     * Temperature controls how much randomness and expressiveness the TTS model uses while generating speech.
+     * Lower values produce more stable and consistent output,
+     * while higher values sound more expressive but may introduce artifacts or errors.
+     *
+     * Any number inbetween 0.01 - 2
+     * @default 0.6
+     *
+     * Note: This parameter is only supported for bulbul:v3. It has no effect on bulbul:v2.
+     */
+    temperature?: number;
+    /**
+     * The ID of a pronunciation dictionary to apply during synthesis.
+     * When provided, matching words in the input text will be replaced with their custom pronunciations before generating speech.
+     *
+     * Only supported by bulbul:v3.
+     */
+    dict_id?: string;
+    /**
+     * Enable caching for the request. When enabled, identical requests will return cached audio instead of regenerating.
+     *
+     * @default false
+     *
+     * Currently in beta and only available for bulbul:v1 and bulbul:v2 models.
+     */
+    enable_cached_responses?: boolean;
 };
 
 type SarvamTranscriptionModelId = "saaras:v3" | "saarika:v2.5" | (string & {});

package/dist/index.d.ts

@@ -3,34 +3,10 @@ import { FetchFunction } from '@ai-sdk/provider-utils';
 import { z } from 'zod';
 
 /**
- * @description Product models
+ * @description Production models
  */
-type SarvamChatModelId = "sarvam-30b" | "sarvam-30b-16k" | "sarvam-105b" | "sarvam-105b-32k" | SarvamChatLegacyModelId | (string & {});
-/**
- * @description Legacy models
- * @deprecated
- */
-type SarvamChatLegacyModelId = "sarvam-m";
+type SarvamChatModelId = "sarvam-30b" | "sarvam-30b-16k" | "sarvam-105b" | "sarvam-105b-32k" | (string & {});
 interface SarvamChatSettings {
-    /**
-    * Whether to simulate artificial tool calling or JSON object generation, because Sarvam Models doen't support native Tool Calling or JSON Schmea.
-    * @default undefined
-    * @example
-        await generateText({
-            model: sarvam("sarvam-m", {
-                simulate: "tool-calling"
-            })
-            tools: {...}
-        })
-
-        await generateObject({
-            model: sarvam("sarvam-m", {
-                simulate: "json-object"
-            })
-            schema: {...}
-        })
-    */
-    simulate?: "tool-calling" | "json-object";
     /**
      * Whether to enable parallel function calling during tool use.
      * @default true
@@ -52,7 +28,7 @@ declare const SarvamLanguageCodeSchema: z.ZodEnum<["hi-IN", "bn-IN", "kn-IN", "m
 
 type SarvamSpeechModelId = "bulbul:v2" | "bulbul:v3" | (string & {});
 type SarvamSpeechVoices = z.infer<typeof SpeakerSchema>;
-declare const SpeakerSchema: z.ZodDefault<z.ZodEnum<["abhilash", "karun", "hitesh", "anushka", "manisha", "vidya", "arya", "shubh", "aditya", "rahul", "rohan", "amit", "dev", "ratan", "varun", "manan", "sumit", "kabir", "aayan", "ashutosh", "advait", "anand", "tarun", "sunny", "mani", "gokul", "vijay", "mohit", "rehan", "soham", "ritu", "priya", "neha", "pooja", "simran", "kavya", "ishita", "shreya", "roopa", "amelia", "sophia", "tanya", "shruti", "suhani", "kavitha", "rupali"]>>;
+declare const SpeakerSchema: z.ZodEnum<["abhilash", "karun", "hitesh", "anushka", "manisha", "vidya", "arya", "shubh", "aditya", "rahul", "rohan", "amit", "dev", "ratan", "varun", "manan", "sumit", "kabir", "aayan", "ashutosh", "advait", "anand", "tarun", "sunny", "mani", "gokul", "vijay", "mohit", "rehan", "soham", "ritu", "priya", "neha", "pooja", "simran", "kavya", "ishita", "shreya", "roopa", "amelia", "sophia", "tanya", "shruti", "suhani", "kavitha", "rupali"]>;
 /**
  * Configuration settings for Sarvam Text-to-Speech API.
  *
@@ -111,6 +87,37 @@ type SarvamSpeechSettings = {
      * @example false (Disable preprocessing)
      */
     enable_preprocessing?: boolean;
+    /**
+     * Specifies the audio codec for the output audio file.
+     * Different codecs offer various compression and quality characteristics.
+     */
+    output_audio_codec?: "mp3" | "linear16" | "mulaw" | "alaw" | "opus" | "flac" | "aac" | "wav";
+    /**
+     * Temperature controls how much randomness and expressiveness the TTS model uses while generating speech.
+     * Lower values produce more stable and consistent output,
+     * while higher values sound more expressive but may introduce artifacts or errors.
+     *
+     * Any number inbetween 0.01 - 2
+     * @default 0.6
+     *
+     * Note: This parameter is only supported for bulbul:v3. It has no effect on bulbul:v2.
+     */
+    temperature?: number;
+    /**
+     * The ID of a pronunciation dictionary to apply during synthesis.
+     * When provided, matching words in the input text will be replaced with their custom pronunciations before generating speech.
+     *
+     * Only supported by bulbul:v3.
+     */
+    dict_id?: string;
+    /**
+     * Enable caching for the request. When enabled, identical requests will return cached audio instead of regenerating.
+     *
+     * @default false
+     *
+     * Currently in beta and only available for bulbul:v1 and bulbul:v2 models.
+     */
+    enable_cached_responses?: boolean;
 };
 
 type SarvamTranscriptionModelId = "saaras:v3" | "saarika:v2.5" | (string & {});