Audio

Text-to-Speech

client.audio.synthesize sends POST /v1/audio/speech and returns a Uint8Array of raw audio bytes.

1 import { MeshAPI } from "meshapi-node-sdk";
2 import { writeFileSync } from "fs";
3 
4 const client = new MeshAPI({ baseUrl: "https://api.meshapi.ai", token: "rsk_..." });
5 
6 const audio = await client.audio.synthesize({
7   input: "Hello from MeshAPI.",
8   model: "sarvam/bulbul:v2",
9   voice: "meera",
10 });
11 
12 writeFileSync("output.wav", Buffer.from(audio));

`SpeechParams` fields

Field	Type	Notes
`input`	`string`	Required. Text to synthesize.
`model`	`string`	Required. e.g. `"sarvam/bulbul:v2"`
`voice`	`string?`	Voice ID or name
`response_format`	`string?`	Audio format, e.g. `"wav"`, `"mp3"`
`speed`	`number?`	Playback speed multiplier

Speech-to-Text (Transcription)

client.audio.transcribe sends POST /v1/audio/transcriptions as a multipart upload and returns a TranscriptionResponse.

1 import { readFileSync } from "fs";
2 
3 const fileBytes = readFileSync("audio.wav");
4 
5 const result = await client.audio.transcribe(
6   fileBytes,
7   { model: "sarvam/saaras:v3" },
8   { filename: "audio.wav" },
9 );
10 
11 console.log(result.text);

`TranscriptionParams` key fields

Field	Type	Notes
`model`	`string`	Required. e.g. `"sarvam/saaras:v3"`
`language_code`	`string?`	Optional. Model-specific language code (e.g. Sarvam expects `"en-IN"`)
`diarize`	`boolean?`	Enable speaker diarization
`num_speakers`	`number?`	Expected number of speakers
`timestamps_granularity`	`string?`	e.g. `"word"` for word-level timestamps
`tag_audio_events`	`boolean?`	Tag non-speech audio events
`additional_formats`	`string?`	Request extra output formats

Transcribe & Translate

client.audio.translate sends POST /v1/audio/transcriptions/translate — it transcribes the audio and translates the transcript to English in one step. Same three-argument shape as transcribe.

1 import { readFileSync } from "fs";
2 
3 const audio = readFileSync("french_audio.mp3");
4 
5 const result = await client.audio.translate(
6   audio,
7   { model: "sarvam/saaras:v3" },
8   { filename: "french_audio.mp3" },
9 );
10 
11 console.log(result.text);

Translation (to English)

client.audio.translations sends POST /v1/audio/translations — a distinct endpoint that transcribes audio in any language and returns the text translated to English. It takes the same three-argument shape as transcribe and returns a TranscriptionResponse (.text is the English translation).

1 import { readFileSync } from "fs";
2 
3 const audio = readFileSync("french_audio.mp3");
4 
5 const result = await client.audio.translations(
6   audio,
7   { model: "openai/whisper-large-v3" },
8   { filename: "french_audio.mp3" },
9 );
10 
11 console.log(result.text); // English translation

Use a speech model that supports translation — see the Models list. model is required.

Optional parameters (passed in the second argument object): prompt (context hint), response_format ("json", "text", or "verbose_json"), and temperature (0–2).

List Voices

client.audio.listVoices sends GET /v1/audio/voices.

1 const voices = await client.audio.listVoices({ page_size: 10 });
2 console.log(voices);

`ListVoicesParams` fields

Field	Type	Notes
`page_size`	`number?`	Results per page
`next_page_token`	`string?`	Pagination cursor
`search`	`string?`	Filter by name
`voice_type`	`string?`	`"standard"`, `"cloned"`, etc.
`category`	`string?`	Voice category filter

Get Voice

client.audio.getVoice sends GET /v1/audio/voices/{voice_id}.

1 const voice = await client.audio.getVoice("voice-id");
2 console.log(voice);