undefined
Class: MoonshineModel
Defined in: model.ts:15
Implements speech-to-text inferences with Moonshine models.
Constructors
new MoonshineModel()
new MoonshineModel(modelURL, precision): MoonshineModelDefined in: model.ts:33
Create (but do not load) a new MoonshineModel for inference.
Parameters
| Parameter | Type | Default value | Description |
|---|---|---|---|
modelURL |
string |
undefined |
A string (relative to Settings.BASE_ASSET_PATH) where the .onnx model weights are located. |
precision |
string |
"quantized" |
- |
Returns
Remarks
Creating a MoonshineModel has the side effect of setting the path to the onnxruntime-web .wasm to the Settings.BASE_ASSET_PATH
Methods
benchmarkExecutionEnvironment()
benchmarkExecutionEnvironment(sampleSize): Promise<number>Defined in: model.ts:100
Tests the inference latency of the current environment. This is useful for determining the appropriate Settings.FRAME_SIZE and Settings.MAX_SPEECH_SECS for a given execution environment.
Parameters
| Parameter | Type | Default value | Description |
|---|---|---|---|
sampleSize |
number |
10 |
(Optional) The number of samples to use for computing the benchmark |
Returns
Promise<number>
The average inference latency (in ms) over the number of samples taken.
Remarks
Warning: since this uses noise to benchmark the model, the model will have lower performance if you to use it for transcription immediately after benchmarking.
generate()
generate(audio): Promise<string>Defined in: model.ts:158
Generate a transcription of the passed audio.
Parameters
| Parameter | Type | Description |
|---|---|---|
audio |
Float32Array |
A Float32Array containing raw audio samples from an audio source (e.g., a wav file, or a user’s microphone) |
Returns
Promise<string>
A Promise that resolves with the generated transcription.
getLatency()
getLatency(): numberDefined in: model.ts:120
Returns the latency (in ms) of the most recent call to MoonshineModel.generate
Returns
number
A latency reading (in ms)
isLoaded()
isLoaded(): booleanDefined in: model.ts:149
Returns whether or not the model weights have been loaded.
Returns
boolean
true if the model is loaded, false if not.
loadModel()
loadModel(): Promise<void>Defined in: model.ts:127
Load the model weights.
Returns
Promise<void>