runanywhere/sherpa-onnx-whisper-tiny.en

Sherpa-ONNX Whisper Tiny — English

OpenAI's Whisper Tiny model converted to ONNX format for sherpa-onnx, packaged for use with the RunAnywhere SDK.

Format: tar.gz archive (~146 MB) containing encoder + decoder ONNX files + tokens

Usage with RunAnywhere SDK

Swift (iOS / macOS)

import RunAnywhere

RunAnywhere.registerModel(
    id: "sherpa-onnx-whisper-tiny.en",
    name: "Whisper Tiny English (ONNX)",
    url: URL(string: "https://huggingface.co/runanywhere/sherpa-onnx-whisper-tiny.en/resolve/main/sherpa-onnx-whisper-tiny.en.tar.gz")!,
    framework: .onnx,
    modality: .speechRecognition,
    artifactType: .archive(.tarGz, structure: .nestedDirectory),
    memoryRequirement: 75_000_000
)

// Transcribe audio
let result = try await RunAnywhere.transcribe(audioData, modelId: "sherpa-onnx-whisper-tiny.en")
print(result.text)

Kotlin (Android / JVM)

import com.runanywhere.sdk.RunAnywhere
import com.runanywhere.sdk.models.*

RunAnywhere.registerModel(
    id = "sherpa-onnx-whisper-tiny.en",
    name = "Whisper Tiny English (ONNX)",
    url = "https://huggingface.co/runanywhere/sherpa-onnx-whisper-tiny.en/resolve/main/sherpa-onnx-whisper-tiny.en.tar.gz",
    framework = InferenceFramework.ONNX,
    modality = ModelCategory.SPEECH_RECOGNITION,
    memoryRequirement = 75_000_000L
)

val result = RunAnywhere.transcribe(audioData, modelId = "sherpa-onnx-whisper-tiny.en")
println(result.text)

Web (TypeScript)

import { RunAnywhere, LLMFramework, ModelCategory } from '@anthropic/runanywhere-web';

RunAnywhere.registerModels([{
  id: 'sherpa-onnx-whisper-tiny.en',
  name: 'Whisper Tiny English (ONNX)',
  url: 'https://huggingface.co/runanywhere/sherpa-onnx-whisper-tiny.en/resolve/main/sherpa-onnx-whisper-tiny.en.tar.gz',
  framework: LLMFramework.ONNX,
  modality: ModelCategory.SpeechRecognition,
  memoryRequirement: 75_000_000,
  artifactType: 'archive',
}]);

await RunAnywhere.downloadModel('sherpa-onnx-whisper-tiny.en');
await RunAnywhere.loadModel('sherpa-onnx-whisper-tiny.en');

const result = await RunAnywhere.transcribe(audioData, 'sherpa-onnx-whisper-tiny.en');
console.log(result.text);

React Native (TypeScript)

import { RunAnywhere } from 'runanywhere-react-native';

RunAnywhere.registerModel({
  id: 'sherpa-onnx-whisper-tiny.en',
  name: 'Whisper Tiny English (ONNX)',
  url: 'https://huggingface.co/runanywhere/sherpa-onnx-whisper-tiny.en/resolve/main/sherpa-onnx-whisper-tiny.en.tar.gz',
  framework: 'onnx',
  modality: 'speechRecognition',
  memoryRequirement: 75_000_000,
});

const result = await RunAnywhere.transcribe(audioData, 'sherpa-onnx-whisper-tiny.en');

Flutter (Dart)

import 'package:runanywhere_flutter/runanywhere_flutter.dart';

RunAnywhere.registerModel(
  id: 'sherpa-onnx-whisper-tiny.en',
  name: 'Whisper Tiny English (ONNX)',
  url: 'https://huggingface.co/runanywhere/sherpa-onnx-whisper-tiny.en/resolve/main/sherpa-onnx-whisper-tiny.en.tar.gz',
  framework: InferenceFramework.onnx,
  modality: ModelCategory.speechRecognition,
  memoryRequirement: 75000000,
);

final result = await RunAnywhere.transcribe(audioData, 'sherpa-onnx-whisper-tiny.en');
print(result.text);

Model Details

Property	Value
Base Model	OpenAI Whisper Tiny
Language	English
Parameters	39M
Format	ONNX (encoder + decoder)
Runtime	sherpa-onnx

Attribution

Based on OpenAI Whisper. ONNX conversion by csukuangfj/sherpa-onnx.