decibri 5.4.0 → 5.5.0
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 +26 -0
- package/README.md +10 -2
- package/examples/decibri.browser.js +3 -1
- package/index.d.ts +11 -0
- package/index.js +52 -52
- package/models/README.md +21 -103
- package/models/THIRD-PARTY-NOTICES.md +110 -0
- package/package.json +5 -5
- package/src/browser/decibri-browser.js +14 -1
- package/src/browser/decibri-output-browser.js +6 -0
- package/src/browser/index.d.ts +5 -2
- package/src/decibri-output.js +6 -2
- package/src/decibri.d.ts +40 -6
- package/src/decibri.js +25 -7
- package/src/errors.js +3 -1
package/CHANGELOG.md
CHANGED
|
@@ -9,6 +9,32 @@ For other decibri packages, see:
|
|
|
9
9
|
- Rust core: [crates/decibri/CHANGELOG.md](../../crates/decibri/CHANGELOG.md)
|
|
10
10
|
- Python package: [bindings/python/CHANGELOG.md](../../bindings/python/CHANGELOG.md)
|
|
11
11
|
|
|
12
|
+
## [5.5.0] - 2026-08-10
|
|
13
|
+
|
|
14
|
+
### Breaking changes
|
|
15
|
+
|
|
16
|
+
- **The `RangeError` thrown for a `channels` count below 1 carries the message `channels must be at least 1`.** It carried `channels must be between 1 and 32`, naming an upper bound that neither `Microphone` nor `Speaker` enforces. Code that matches the message text exactly stops matching and has to be updated to the new text. Nothing else about the error moves: it is still a `RangeError`, thrown for the same input, `channels: 0` on either constructor or `open()`.
|
|
17
|
+
- **`Speaker` accepts a `channels` count above 32.** It threw `RangeError: channels must be between 1 and 32` from the constructor. It now accepts any count above zero, offers the count to the device when playback starts, and a device that cannot serve it emits a `DecibriError` with code `SPEAKER_CHANNELS_UNSUPPORTED` on the `'error'` event, or rejects `writeAsync()` with it. Code that relied on the constructor throwing to reject an unsupported output channel count has to handle the failure asynchronously instead, on `'error'` or the rejection. `new Speaker({ channels: 0 })` still throws a `RangeError` from the constructor, and a count above 16383 still carries `STREAM_OPEN_FAILED` naming that limit. How many channels a device serves depends on the `sampleRate` asked for as well as the count.
|
|
18
|
+
- **An output channel count above the device's reported figure carries `SPEAKER_CHANNELS_UNSUPPORTED`.** It carried `STREAM_OPEN_FAILED`. Code branching on `STREAM_OPEN_FAILED` to detect an unsupported output channel count has to match `SPEAKER_CHANNELS_UNSUPPORTED` as well, and cannot drop `STREAM_OPEN_FAILED`, because a device that reports no figure at all keeps `STREAM_OPEN_FAILED` for the same condition. `STREAM_OPEN_FAILED` is otherwise unchanged and remains the code for an output open that failed for any other reason.
|
|
19
|
+
- **The browser `Microphone` rejects a `channels` count above 1.** It accepted 1 to 32 and delivered the first channel only, with nothing to indicate the rest were dropped. A count above 1 now throws `RangeError: multichannel capture is not supported; channels must be 1 (mono)`, and a count below 1 throws `RangeError: channels must be at least 1` where it threw a `TypeError`. Code passing a count above 1 to the browser entry has to pass 1 and mix down its own sources, or read the single channel it was already receiving. `channels: 1` is unchanged, as is every other browser `Microphone` option. Both class and message now match the Node entry's for the same values, so the two entries reject the same input the same way.
|
|
20
|
+
- The browser `Speaker` is unchanged. It still accepts 1 to 32 channels, the range the Web Audio specification requires an implementation to support, and keeps its own `channels must be between 1 and 32, got <n>` message, which that range does enforce.
|
|
21
|
+
|
|
22
|
+
### Added
|
|
23
|
+
|
|
24
|
+
- Error code `SPEAKER_CHANNELS_UNSUPPORTED` on `DecibriError`, for an output device that cannot serve the requested `channels`. The message names the count asked for, the count the device reports (the figure `SpeakerInfo.maxOutputChannels` carries) and the platform's own message.
|
|
25
|
+
- `referenceChannels` on the `aec` option object: the channel count of the far-end reference pushed through `pushAecReference`. Default 1 (mono). With a count above 1 the pushed samples are read as interleaved frames and each frame is averaged to one mono sample before the canceller sees it. The collapse is opt-in: a caller pushing a multichannel reference must declare the count, and an undeclared multichannel push keeps its current behaviour, cancelling nothing and reporting no error. The declared count must match the pushed buffer: a mismatch is not detected and raises no error, and shows up only as `aecMetrics().delaySamples` staying `null` with no fault reported. A count below 1 throws a `RangeError`; the only ceiling is the option's own 16-bit carrier. A mono reference against playback through more than one loudspeaker has a cancellation ceiling: the canceller models one room response applied to the channel average, so a placement where the per-loudspeaker echo paths differ leaves a residual that adaptation does not remove.
|
|
26
|
+
- Each of the four platform packages (`@decibri/decibri-win32-x64-msvc`, `@decibri/decibri-darwin-arm64`, `@decibri/decibri-linux-x64-gnu`, `@decibri/decibri-linux-arm64-gnu`) now includes `THIRD-PARTY-NOTICES.md`, the third-party license notices for the ONNX Runtime dynamic libraries the package carries and for the third-party material incorporated into them, with a source-availability statement for the MPL-2.0-licensed Eigen code they contain.
|
|
27
|
+
- `models/THIRD-PARTY-NOTICES.md`, carrying the origin, version and license text for the two bundled ONNX models together with the training-data attribution the denoise checkpoint requires. It ships beside the weights it covers. `models/README.md` alongside it documents each model's tensor interface and points at the notice.
|
|
28
|
+
|
|
29
|
+
### Changed
|
|
30
|
+
|
|
31
|
+
- ONNX Runtime telemetry is disabled on the environment decibri commits when it initializes the runtime, where it was left at ONNX Runtime's own default of enabled. This covers every path that reaches the runtime: `vad: 'silero'` and the ACE `denoise` stage. Set `DECIBRI_ORT_TELEMETRY=1` in the environment before first use to leave it enabled; every other value, an empty value, and an absent variable leave it disabled. Two limits apply on Windows and decibri can close neither. ONNX Runtime logs one process-information event while the environment is being created, before the setting is applied, and logs it once per process, so that event is emitted whichever way the setting is left. The runtime also assigns its telemetry state from the Windows tracing session through an ETW callback, so the platform can re-enable telemetry after decibri has disabled it. On other platforms ONNX Runtime's telemetry provider does nothing. The browser build has no ONNX Runtime and is unaffected. No option, event, method or error text changes.
|
|
32
|
+
|
|
33
|
+
### Fixed
|
|
34
|
+
|
|
35
|
+
- The bundled Silero VAD model is documented as v6.2, the version that ships. The `model` field on `VadOptions`, the `vad` option on `MicrophoneOptions`, the README and the notice beside the model named v5. Which model file ships is unchanged.
|
|
36
|
+
- The Silero VAD tensor specification in `models/README.md` names the tensors the model exposes: `input`, `state` and `sr` in, `output` and `stateN` out. It described a four-input form carrying separate `h` and `c` LSTM tensors.
|
|
37
|
+
|
|
12
38
|
## [5.4.0] - 2026-08-04
|
|
13
39
|
|
|
14
40
|
### Added
|
package/README.md
CHANGED
|
@@ -152,7 +152,7 @@ Creates a Writable stream for speaker playback.
|
|
|
152
152
|
| Option | Type | Default | Description |
|
|
153
153
|
| --- | --- | --- | --- |
|
|
154
154
|
| `sampleRate` | number | 16000 | Playback sample rate (1000 to 384000) |
|
|
155
|
-
| `channels` | number | 1 | Output channels (1 to
|
|
155
|
+
| `channels` | number | 1 | Output channels (1 or more, up to what the device supports) |
|
|
156
156
|
| `dtype` | `'int16'` \| `'float32'` | `'int16'` | Sample encoding of incoming data |
|
|
157
157
|
| `device` | number, string, or `{ id: string }` | system default | Output device index, case-insensitive name substring, or stable per-host ID |
|
|
158
158
|
|
|
@@ -293,7 +293,7 @@ mic.on('silence', () => console.log('silent'));
|
|
|
293
293
|
|
|
294
294
|
### Silero mode
|
|
295
295
|
|
|
296
|
-
ML-based detection using the Silero VAD
|
|
296
|
+
ML-based detection using the Silero VAD v6.2 model. More accurate than energy mode, especially in noisy environments.
|
|
297
297
|
|
|
298
298
|
```javascript
|
|
299
299
|
const mic = new Microphone({ vad: { model: 'silero', threshold: 0.5 } });
|
|
@@ -337,6 +337,14 @@ setTimeout(() => mic.stop(), 5000);
|
|
|
337
337
|
|
|
338
338
|
VAD reads the signal before the chain, so `vadScore` and the `'speech'` / `'silence'` events are unaffected by which conditioning stages you enable. The conditioning chain runs in the native Node.js capture path; the browser build does not include it.
|
|
339
339
|
|
|
340
|
+
## ONNX Runtime telemetry
|
|
341
|
+
|
|
342
|
+
Silero mode (`vad: 'silero'`) and the ACE `denoise` stage run on ONNX Runtime, which carries its own telemetry, separate from anything decibri does. Decibri disables it on the environment it commits when it initializes the runtime. Set `DECIBRI_ORT_TELEMETRY=1` in the environment before first use to leave it enabled; every other value, an empty value, and an absent variable leave it disabled.
|
|
343
|
+
|
|
344
|
+
Two limits apply on Windows and decibri can close neither, so decibri does not claim that no telemetry is emitted. ONNX Runtime logs one process-information event while the environment is being created, before the setting is applied, and logs it once per process, so that event is emitted whichever way the setting is left. The runtime also assigns its telemetry state from the Windows tracing session through an ETW callback, so the platform can re-enable telemetry after decibri has disabled it. On other platforms ONNX Runtime's telemetry provider does nothing.
|
|
345
|
+
|
|
346
|
+
Neither ONNX Runtime nor this setting applies to the browser build, which has no ONNX Runtime.
|
|
347
|
+
|
|
340
348
|
## API: File (offline source)
|
|
341
349
|
|
|
342
350
|
Everything a `Microphone` does to live audio, `File` does to audio you already have: the same conditioning options, the same Readable stream of conditioned chunks (finite: it ends at EOF), and the same opt-in `vad`. Because a `File` is a complete recording, it can also analyze the whole recording for speech.
|
|
@@ -70,7 +70,7 @@ var decibri = (function() {
|
|
|
70
70
|
var require_decibri_browser = /* @__PURE__ */ __commonJSMin(((exports, module) => {
|
|
71
71
|
const { Emitter } = require_emitter();
|
|
72
72
|
const { WORKLET_SOURCE } = require_worklet_inline();
|
|
73
|
-
const VERSION = "5.
|
|
73
|
+
const VERSION = "5.5.0";
|
|
74
74
|
/**
|
|
75
75
|
* Browser microphone capture.
|
|
76
76
|
*
|
|
@@ -132,6 +132,8 @@ var decibri = (function() {
|
|
|
132
132
|
this._noiseSuppression = options.noiseSuppression ?? true;
|
|
133
133
|
this._workletUrl = options.workletUrl;
|
|
134
134
|
if (this._sampleRate < 1e3 || this._sampleRate > 384e3) throw new RangeError("sample rate must be between 1000 and 384000");
|
|
135
|
+
if (this._channels < 1) throw new RangeError("channels must be at least 1");
|
|
136
|
+
if (this._channels > 1) throw new RangeError("multichannel capture is not supported; channels must be 1 (mono)");
|
|
135
137
|
if (this._channels < 1 || this._channels > 32) throw new TypeError(`channels must be between 1 and 32, got ${this._channels}`);
|
|
136
138
|
if (this._framesPerBuffer < 64 || this._framesPerBuffer > 65536) throw new TypeError(`frames per buffer must be between 64 and 65536, got ${this._framesPerBuffer}`);
|
|
137
139
|
if (this._dtype !== "int16" && this._dtype !== "float32") throw new TypeError("dtype must be 'int16' or 'float32'");
|
package/index.d.ts
CHANGED
|
@@ -324,6 +324,17 @@ export interface DecibriOptions {
|
|
|
324
324
|
* when `aec` names a model.
|
|
325
325
|
*/
|
|
326
326
|
aecReferenceSampleRate?: number
|
|
327
|
+
/**
|
|
328
|
+
* Number of channels in the far-end reference pushed through
|
|
329
|
+
* `pushAecReference`, at least 1. Absent means the reference is mono;
|
|
330
|
+
* when it names a count above 1, the core collapses each interleaved
|
|
331
|
+
* frame to one mono sample before the canceller sees it. The declared
|
|
332
|
+
* count must match the pushed buffer: a mismatch is not detected and
|
|
333
|
+
* raises no error, and shows up only as `aecMetrics().delaySamples`
|
|
334
|
+
* staying `null` with no fault reported. Consulted only when `aec`
|
|
335
|
+
* names a model.
|
|
336
|
+
*/
|
|
337
|
+
aecReferenceChannels?: number
|
|
327
338
|
}
|
|
328
339
|
|
|
329
340
|
/** Options passed from JS constructor for output. */
|
package/index.js
CHANGED
|
@@ -77,8 +77,8 @@ function requireNative() {
|
|
|
77
77
|
try {
|
|
78
78
|
const binding = require('@decibri/decibri-android-arm64')
|
|
79
79
|
const bindingPackageVersion = require('@decibri/decibri-android-arm64/package.json').version
|
|
80
|
-
if (bindingPackageVersion !== '5.
|
|
81
|
-
throw new Error(`Native binding package version mismatch, expected 5.
|
|
80
|
+
if (bindingPackageVersion !== '5.5.0' && process.env.NAPI_RS_ENFORCE_VERSION_CHECK && process.env.NAPI_RS_ENFORCE_VERSION_CHECK !== '0') {
|
|
81
|
+
throw new Error(`Native binding package version mismatch, expected 5.5.0 but got ${bindingPackageVersion}. You can reinstall dependencies to fix this issue.`)
|
|
82
82
|
}
|
|
83
83
|
return binding
|
|
84
84
|
} catch (e) {
|
|
@@ -93,8 +93,8 @@ function requireNative() {
|
|
|
93
93
|
try {
|
|
94
94
|
const binding = require('@decibri/decibri-android-arm-eabi')
|
|
95
95
|
const bindingPackageVersion = require('@decibri/decibri-android-arm-eabi/package.json').version
|
|
96
|
-
if (bindingPackageVersion !== '5.
|
|
97
|
-
throw new Error(`Native binding package version mismatch, expected 5.
|
|
96
|
+
if (bindingPackageVersion !== '5.5.0' && process.env.NAPI_RS_ENFORCE_VERSION_CHECK && process.env.NAPI_RS_ENFORCE_VERSION_CHECK !== '0') {
|
|
97
|
+
throw new Error(`Native binding package version mismatch, expected 5.5.0 but got ${bindingPackageVersion}. You can reinstall dependencies to fix this issue.`)
|
|
98
98
|
}
|
|
99
99
|
return binding
|
|
100
100
|
} catch (e) {
|
|
@@ -114,8 +114,8 @@ function requireNative() {
|
|
|
114
114
|
try {
|
|
115
115
|
const binding = require('@decibri/decibri-win32-x64-gnu')
|
|
116
116
|
const bindingPackageVersion = require('@decibri/decibri-win32-x64-gnu/package.json').version
|
|
117
|
-
if (bindingPackageVersion !== '5.
|
|
118
|
-
throw new Error(`Native binding package version mismatch, expected 5.
|
|
117
|
+
if (bindingPackageVersion !== '5.5.0' && process.env.NAPI_RS_ENFORCE_VERSION_CHECK && process.env.NAPI_RS_ENFORCE_VERSION_CHECK !== '0') {
|
|
118
|
+
throw new Error(`Native binding package version mismatch, expected 5.5.0 but got ${bindingPackageVersion}. You can reinstall dependencies to fix this issue.`)
|
|
119
119
|
}
|
|
120
120
|
return binding
|
|
121
121
|
} catch (e) {
|
|
@@ -130,8 +130,8 @@ function requireNative() {
|
|
|
130
130
|
try {
|
|
131
131
|
const binding = require('@decibri/decibri-win32-x64-msvc')
|
|
132
132
|
const bindingPackageVersion = require('@decibri/decibri-win32-x64-msvc/package.json').version
|
|
133
|
-
if (bindingPackageVersion !== '5.
|
|
134
|
-
throw new Error(`Native binding package version mismatch, expected 5.
|
|
133
|
+
if (bindingPackageVersion !== '5.5.0' && process.env.NAPI_RS_ENFORCE_VERSION_CHECK && process.env.NAPI_RS_ENFORCE_VERSION_CHECK !== '0') {
|
|
134
|
+
throw new Error(`Native binding package version mismatch, expected 5.5.0 but got ${bindingPackageVersion}. You can reinstall dependencies to fix this issue.`)
|
|
135
135
|
}
|
|
136
136
|
return binding
|
|
137
137
|
} catch (e) {
|
|
@@ -147,8 +147,8 @@ function requireNative() {
|
|
|
147
147
|
try {
|
|
148
148
|
const binding = require('@decibri/decibri-win32-ia32-msvc')
|
|
149
149
|
const bindingPackageVersion = require('@decibri/decibri-win32-ia32-msvc/package.json').version
|
|
150
|
-
if (bindingPackageVersion !== '5.
|
|
151
|
-
throw new Error(`Native binding package version mismatch, expected 5.
|
|
150
|
+
if (bindingPackageVersion !== '5.5.0' && process.env.NAPI_RS_ENFORCE_VERSION_CHECK && process.env.NAPI_RS_ENFORCE_VERSION_CHECK !== '0') {
|
|
151
|
+
throw new Error(`Native binding package version mismatch, expected 5.5.0 but got ${bindingPackageVersion}. You can reinstall dependencies to fix this issue.`)
|
|
152
152
|
}
|
|
153
153
|
return binding
|
|
154
154
|
} catch (e) {
|
|
@@ -163,8 +163,8 @@ function requireNative() {
|
|
|
163
163
|
try {
|
|
164
164
|
const binding = require('@decibri/decibri-win32-arm64-msvc')
|
|
165
165
|
const bindingPackageVersion = require('@decibri/decibri-win32-arm64-msvc/package.json').version
|
|
166
|
-
if (bindingPackageVersion !== '5.
|
|
167
|
-
throw new Error(`Native binding package version mismatch, expected 5.
|
|
166
|
+
if (bindingPackageVersion !== '5.5.0' && process.env.NAPI_RS_ENFORCE_VERSION_CHECK && process.env.NAPI_RS_ENFORCE_VERSION_CHECK !== '0') {
|
|
167
|
+
throw new Error(`Native binding package version mismatch, expected 5.5.0 but got ${bindingPackageVersion}. You can reinstall dependencies to fix this issue.`)
|
|
168
168
|
}
|
|
169
169
|
return binding
|
|
170
170
|
} catch (e) {
|
|
@@ -182,8 +182,8 @@ function requireNative() {
|
|
|
182
182
|
try {
|
|
183
183
|
const binding = require('@decibri/decibri-darwin-universal')
|
|
184
184
|
const bindingPackageVersion = require('@decibri/decibri-darwin-universal/package.json').version
|
|
185
|
-
if (bindingPackageVersion !== '5.
|
|
186
|
-
throw new Error(`Native binding package version mismatch, expected 5.
|
|
185
|
+
if (bindingPackageVersion !== '5.5.0' && process.env.NAPI_RS_ENFORCE_VERSION_CHECK && process.env.NAPI_RS_ENFORCE_VERSION_CHECK !== '0') {
|
|
186
|
+
throw new Error(`Native binding package version mismatch, expected 5.5.0 but got ${bindingPackageVersion}. You can reinstall dependencies to fix this issue.`)
|
|
187
187
|
}
|
|
188
188
|
return binding
|
|
189
189
|
} catch (e) {
|
|
@@ -198,8 +198,8 @@ function requireNative() {
|
|
|
198
198
|
try {
|
|
199
199
|
const binding = require('@decibri/decibri-darwin-x64')
|
|
200
200
|
const bindingPackageVersion = require('@decibri/decibri-darwin-x64/package.json').version
|
|
201
|
-
if (bindingPackageVersion !== '5.
|
|
202
|
-
throw new Error(`Native binding package version mismatch, expected 5.
|
|
201
|
+
if (bindingPackageVersion !== '5.5.0' && process.env.NAPI_RS_ENFORCE_VERSION_CHECK && process.env.NAPI_RS_ENFORCE_VERSION_CHECK !== '0') {
|
|
202
|
+
throw new Error(`Native binding package version mismatch, expected 5.5.0 but got ${bindingPackageVersion}. You can reinstall dependencies to fix this issue.`)
|
|
203
203
|
}
|
|
204
204
|
return binding
|
|
205
205
|
} catch (e) {
|
|
@@ -214,8 +214,8 @@ function requireNative() {
|
|
|
214
214
|
try {
|
|
215
215
|
const binding = require('@decibri/decibri-darwin-arm64')
|
|
216
216
|
const bindingPackageVersion = require('@decibri/decibri-darwin-arm64/package.json').version
|
|
217
|
-
if (bindingPackageVersion !== '5.
|
|
218
|
-
throw new Error(`Native binding package version mismatch, expected 5.
|
|
217
|
+
if (bindingPackageVersion !== '5.5.0' && process.env.NAPI_RS_ENFORCE_VERSION_CHECK && process.env.NAPI_RS_ENFORCE_VERSION_CHECK !== '0') {
|
|
218
|
+
throw new Error(`Native binding package version mismatch, expected 5.5.0 but got ${bindingPackageVersion}. You can reinstall dependencies to fix this issue.`)
|
|
219
219
|
}
|
|
220
220
|
return binding
|
|
221
221
|
} catch (e) {
|
|
@@ -234,8 +234,8 @@ function requireNative() {
|
|
|
234
234
|
try {
|
|
235
235
|
const binding = require('@decibri/decibri-freebsd-x64')
|
|
236
236
|
const bindingPackageVersion = require('@decibri/decibri-freebsd-x64/package.json').version
|
|
237
|
-
if (bindingPackageVersion !== '5.
|
|
238
|
-
throw new Error(`Native binding package version mismatch, expected 5.
|
|
237
|
+
if (bindingPackageVersion !== '5.5.0' && process.env.NAPI_RS_ENFORCE_VERSION_CHECK && process.env.NAPI_RS_ENFORCE_VERSION_CHECK !== '0') {
|
|
238
|
+
throw new Error(`Native binding package version mismatch, expected 5.5.0 but got ${bindingPackageVersion}. You can reinstall dependencies to fix this issue.`)
|
|
239
239
|
}
|
|
240
240
|
return binding
|
|
241
241
|
} catch (e) {
|
|
@@ -250,8 +250,8 @@ function requireNative() {
|
|
|
250
250
|
try {
|
|
251
251
|
const binding = require('@decibri/decibri-freebsd-arm64')
|
|
252
252
|
const bindingPackageVersion = require('@decibri/decibri-freebsd-arm64/package.json').version
|
|
253
|
-
if (bindingPackageVersion !== '5.
|
|
254
|
-
throw new Error(`Native binding package version mismatch, expected 5.
|
|
253
|
+
if (bindingPackageVersion !== '5.5.0' && process.env.NAPI_RS_ENFORCE_VERSION_CHECK && process.env.NAPI_RS_ENFORCE_VERSION_CHECK !== '0') {
|
|
254
|
+
throw new Error(`Native binding package version mismatch, expected 5.5.0 but got ${bindingPackageVersion}. You can reinstall dependencies to fix this issue.`)
|
|
255
255
|
}
|
|
256
256
|
return binding
|
|
257
257
|
} catch (e) {
|
|
@@ -271,8 +271,8 @@ function requireNative() {
|
|
|
271
271
|
try {
|
|
272
272
|
const binding = require('@decibri/decibri-linux-x64-musl')
|
|
273
273
|
const bindingPackageVersion = require('@decibri/decibri-linux-x64-musl/package.json').version
|
|
274
|
-
if (bindingPackageVersion !== '5.
|
|
275
|
-
throw new Error(`Native binding package version mismatch, expected 5.
|
|
274
|
+
if (bindingPackageVersion !== '5.5.0' && process.env.NAPI_RS_ENFORCE_VERSION_CHECK && process.env.NAPI_RS_ENFORCE_VERSION_CHECK !== '0') {
|
|
275
|
+
throw new Error(`Native binding package version mismatch, expected 5.5.0 but got ${bindingPackageVersion}. You can reinstall dependencies to fix this issue.`)
|
|
276
276
|
}
|
|
277
277
|
return binding
|
|
278
278
|
} catch (e) {
|
|
@@ -287,8 +287,8 @@ function requireNative() {
|
|
|
287
287
|
try {
|
|
288
288
|
const binding = require('@decibri/decibri-linux-x64-gnu')
|
|
289
289
|
const bindingPackageVersion = require('@decibri/decibri-linux-x64-gnu/package.json').version
|
|
290
|
-
if (bindingPackageVersion !== '5.
|
|
291
|
-
throw new Error(`Native binding package version mismatch, expected 5.
|
|
290
|
+
if (bindingPackageVersion !== '5.5.0' && process.env.NAPI_RS_ENFORCE_VERSION_CHECK && process.env.NAPI_RS_ENFORCE_VERSION_CHECK !== '0') {
|
|
291
|
+
throw new Error(`Native binding package version mismatch, expected 5.5.0 but got ${bindingPackageVersion}. You can reinstall dependencies to fix this issue.`)
|
|
292
292
|
}
|
|
293
293
|
return binding
|
|
294
294
|
} catch (e) {
|
|
@@ -305,8 +305,8 @@ function requireNative() {
|
|
|
305
305
|
try {
|
|
306
306
|
const binding = require('@decibri/decibri-linux-arm64-musl')
|
|
307
307
|
const bindingPackageVersion = require('@decibri/decibri-linux-arm64-musl/package.json').version
|
|
308
|
-
if (bindingPackageVersion !== '5.
|
|
309
|
-
throw new Error(`Native binding package version mismatch, expected 5.
|
|
308
|
+
if (bindingPackageVersion !== '5.5.0' && process.env.NAPI_RS_ENFORCE_VERSION_CHECK && process.env.NAPI_RS_ENFORCE_VERSION_CHECK !== '0') {
|
|
309
|
+
throw new Error(`Native binding package version mismatch, expected 5.5.0 but got ${bindingPackageVersion}. You can reinstall dependencies to fix this issue.`)
|
|
310
310
|
}
|
|
311
311
|
return binding
|
|
312
312
|
} catch (e) {
|
|
@@ -321,8 +321,8 @@ function requireNative() {
|
|
|
321
321
|
try {
|
|
322
322
|
const binding = require('@decibri/decibri-linux-arm64-gnu')
|
|
323
323
|
const bindingPackageVersion = require('@decibri/decibri-linux-arm64-gnu/package.json').version
|
|
324
|
-
if (bindingPackageVersion !== '5.
|
|
325
|
-
throw new Error(`Native binding package version mismatch, expected 5.
|
|
324
|
+
if (bindingPackageVersion !== '5.5.0' && process.env.NAPI_RS_ENFORCE_VERSION_CHECK && process.env.NAPI_RS_ENFORCE_VERSION_CHECK !== '0') {
|
|
325
|
+
throw new Error(`Native binding package version mismatch, expected 5.5.0 but got ${bindingPackageVersion}. You can reinstall dependencies to fix this issue.`)
|
|
326
326
|
}
|
|
327
327
|
return binding
|
|
328
328
|
} catch (e) {
|
|
@@ -339,8 +339,8 @@ function requireNative() {
|
|
|
339
339
|
try {
|
|
340
340
|
const binding = require('@decibri/decibri-linux-arm-musleabihf')
|
|
341
341
|
const bindingPackageVersion = require('@decibri/decibri-linux-arm-musleabihf/package.json').version
|
|
342
|
-
if (bindingPackageVersion !== '5.
|
|
343
|
-
throw new Error(`Native binding package version mismatch, expected 5.
|
|
342
|
+
if (bindingPackageVersion !== '5.5.0' && process.env.NAPI_RS_ENFORCE_VERSION_CHECK && process.env.NAPI_RS_ENFORCE_VERSION_CHECK !== '0') {
|
|
343
|
+
throw new Error(`Native binding package version mismatch, expected 5.5.0 but got ${bindingPackageVersion}. You can reinstall dependencies to fix this issue.`)
|
|
344
344
|
}
|
|
345
345
|
return binding
|
|
346
346
|
} catch (e) {
|
|
@@ -355,8 +355,8 @@ function requireNative() {
|
|
|
355
355
|
try {
|
|
356
356
|
const binding = require('@decibri/decibri-linux-arm-gnueabihf')
|
|
357
357
|
const bindingPackageVersion = require('@decibri/decibri-linux-arm-gnueabihf/package.json').version
|
|
358
|
-
if (bindingPackageVersion !== '5.
|
|
359
|
-
throw new Error(`Native binding package version mismatch, expected 5.
|
|
358
|
+
if (bindingPackageVersion !== '5.5.0' && process.env.NAPI_RS_ENFORCE_VERSION_CHECK && process.env.NAPI_RS_ENFORCE_VERSION_CHECK !== '0') {
|
|
359
|
+
throw new Error(`Native binding package version mismatch, expected 5.5.0 but got ${bindingPackageVersion}. You can reinstall dependencies to fix this issue.`)
|
|
360
360
|
}
|
|
361
361
|
return binding
|
|
362
362
|
} catch (e) {
|
|
@@ -373,8 +373,8 @@ function requireNative() {
|
|
|
373
373
|
try {
|
|
374
374
|
const binding = require('@decibri/decibri-linux-loong64-musl')
|
|
375
375
|
const bindingPackageVersion = require('@decibri/decibri-linux-loong64-musl/package.json').version
|
|
376
|
-
if (bindingPackageVersion !== '5.
|
|
377
|
-
throw new Error(`Native binding package version mismatch, expected 5.
|
|
376
|
+
if (bindingPackageVersion !== '5.5.0' && process.env.NAPI_RS_ENFORCE_VERSION_CHECK && process.env.NAPI_RS_ENFORCE_VERSION_CHECK !== '0') {
|
|
377
|
+
throw new Error(`Native binding package version mismatch, expected 5.5.0 but got ${bindingPackageVersion}. You can reinstall dependencies to fix this issue.`)
|
|
378
378
|
}
|
|
379
379
|
return binding
|
|
380
380
|
} catch (e) {
|
|
@@ -389,8 +389,8 @@ function requireNative() {
|
|
|
389
389
|
try {
|
|
390
390
|
const binding = require('@decibri/decibri-linux-loong64-gnu')
|
|
391
391
|
const bindingPackageVersion = require('@decibri/decibri-linux-loong64-gnu/package.json').version
|
|
392
|
-
if (bindingPackageVersion !== '5.
|
|
393
|
-
throw new Error(`Native binding package version mismatch, expected 5.
|
|
392
|
+
if (bindingPackageVersion !== '5.5.0' && process.env.NAPI_RS_ENFORCE_VERSION_CHECK && process.env.NAPI_RS_ENFORCE_VERSION_CHECK !== '0') {
|
|
393
|
+
throw new Error(`Native binding package version mismatch, expected 5.5.0 but got ${bindingPackageVersion}. You can reinstall dependencies to fix this issue.`)
|
|
394
394
|
}
|
|
395
395
|
return binding
|
|
396
396
|
} catch (e) {
|
|
@@ -407,8 +407,8 @@ function requireNative() {
|
|
|
407
407
|
try {
|
|
408
408
|
const binding = require('@decibri/decibri-linux-riscv64-musl')
|
|
409
409
|
const bindingPackageVersion = require('@decibri/decibri-linux-riscv64-musl/package.json').version
|
|
410
|
-
if (bindingPackageVersion !== '5.
|
|
411
|
-
throw new Error(`Native binding package version mismatch, expected 5.
|
|
410
|
+
if (bindingPackageVersion !== '5.5.0' && process.env.NAPI_RS_ENFORCE_VERSION_CHECK && process.env.NAPI_RS_ENFORCE_VERSION_CHECK !== '0') {
|
|
411
|
+
throw new Error(`Native binding package version mismatch, expected 5.5.0 but got ${bindingPackageVersion}. You can reinstall dependencies to fix this issue.`)
|
|
412
412
|
}
|
|
413
413
|
return binding
|
|
414
414
|
} catch (e) {
|
|
@@ -423,8 +423,8 @@ function requireNative() {
|
|
|
423
423
|
try {
|
|
424
424
|
const binding = require('@decibri/decibri-linux-riscv64-gnu')
|
|
425
425
|
const bindingPackageVersion = require('@decibri/decibri-linux-riscv64-gnu/package.json').version
|
|
426
|
-
if (bindingPackageVersion !== '5.
|
|
427
|
-
throw new Error(`Native binding package version mismatch, expected 5.
|
|
426
|
+
if (bindingPackageVersion !== '5.5.0' && process.env.NAPI_RS_ENFORCE_VERSION_CHECK && process.env.NAPI_RS_ENFORCE_VERSION_CHECK !== '0') {
|
|
427
|
+
throw new Error(`Native binding package version mismatch, expected 5.5.0 but got ${bindingPackageVersion}. You can reinstall dependencies to fix this issue.`)
|
|
428
428
|
}
|
|
429
429
|
return binding
|
|
430
430
|
} catch (e) {
|
|
@@ -440,8 +440,8 @@ function requireNative() {
|
|
|
440
440
|
try {
|
|
441
441
|
const binding = require('@decibri/decibri-linux-ppc64-gnu')
|
|
442
442
|
const bindingPackageVersion = require('@decibri/decibri-linux-ppc64-gnu/package.json').version
|
|
443
|
-
if (bindingPackageVersion !== '5.
|
|
444
|
-
throw new Error(`Native binding package version mismatch, expected 5.
|
|
443
|
+
if (bindingPackageVersion !== '5.5.0' && process.env.NAPI_RS_ENFORCE_VERSION_CHECK && process.env.NAPI_RS_ENFORCE_VERSION_CHECK !== '0') {
|
|
444
|
+
throw new Error(`Native binding package version mismatch, expected 5.5.0 but got ${bindingPackageVersion}. You can reinstall dependencies to fix this issue.`)
|
|
445
445
|
}
|
|
446
446
|
return binding
|
|
447
447
|
} catch (e) {
|
|
@@ -456,8 +456,8 @@ function requireNative() {
|
|
|
456
456
|
try {
|
|
457
457
|
const binding = require('@decibri/decibri-linux-s390x-gnu')
|
|
458
458
|
const bindingPackageVersion = require('@decibri/decibri-linux-s390x-gnu/package.json').version
|
|
459
|
-
if (bindingPackageVersion !== '5.
|
|
460
|
-
throw new Error(`Native binding package version mismatch, expected 5.
|
|
459
|
+
if (bindingPackageVersion !== '5.5.0' && process.env.NAPI_RS_ENFORCE_VERSION_CHECK && process.env.NAPI_RS_ENFORCE_VERSION_CHECK !== '0') {
|
|
460
|
+
throw new Error(`Native binding package version mismatch, expected 5.5.0 but got ${bindingPackageVersion}. You can reinstall dependencies to fix this issue.`)
|
|
461
461
|
}
|
|
462
462
|
return binding
|
|
463
463
|
} catch (e) {
|
|
@@ -476,8 +476,8 @@ function requireNative() {
|
|
|
476
476
|
try {
|
|
477
477
|
const binding = require('@decibri/decibri-openharmony-arm64')
|
|
478
478
|
const bindingPackageVersion = require('@decibri/decibri-openharmony-arm64/package.json').version
|
|
479
|
-
if (bindingPackageVersion !== '5.
|
|
480
|
-
throw new Error(`Native binding package version mismatch, expected 5.
|
|
479
|
+
if (bindingPackageVersion !== '5.5.0' && process.env.NAPI_RS_ENFORCE_VERSION_CHECK && process.env.NAPI_RS_ENFORCE_VERSION_CHECK !== '0') {
|
|
480
|
+
throw new Error(`Native binding package version mismatch, expected 5.5.0 but got ${bindingPackageVersion}. You can reinstall dependencies to fix this issue.`)
|
|
481
481
|
}
|
|
482
482
|
return binding
|
|
483
483
|
} catch (e) {
|
|
@@ -492,8 +492,8 @@ function requireNative() {
|
|
|
492
492
|
try {
|
|
493
493
|
const binding = require('@decibri/decibri-openharmony-x64')
|
|
494
494
|
const bindingPackageVersion = require('@decibri/decibri-openharmony-x64/package.json').version
|
|
495
|
-
if (bindingPackageVersion !== '5.
|
|
496
|
-
throw new Error(`Native binding package version mismatch, expected 5.
|
|
495
|
+
if (bindingPackageVersion !== '5.5.0' && process.env.NAPI_RS_ENFORCE_VERSION_CHECK && process.env.NAPI_RS_ENFORCE_VERSION_CHECK !== '0') {
|
|
496
|
+
throw new Error(`Native binding package version mismatch, expected 5.5.0 but got ${bindingPackageVersion}. You can reinstall dependencies to fix this issue.`)
|
|
497
497
|
}
|
|
498
498
|
return binding
|
|
499
499
|
} catch (e) {
|
|
@@ -508,8 +508,8 @@ function requireNative() {
|
|
|
508
508
|
try {
|
|
509
509
|
const binding = require('@decibri/decibri-openharmony-arm')
|
|
510
510
|
const bindingPackageVersion = require('@decibri/decibri-openharmony-arm/package.json').version
|
|
511
|
-
if (bindingPackageVersion !== '5.
|
|
512
|
-
throw new Error(`Native binding package version mismatch, expected 5.
|
|
511
|
+
if (bindingPackageVersion !== '5.5.0' && process.env.NAPI_RS_ENFORCE_VERSION_CHECK && process.env.NAPI_RS_ENFORCE_VERSION_CHECK !== '0') {
|
|
512
|
+
throw new Error(`Native binding package version mismatch, expected 5.5.0 but got ${bindingPackageVersion}. You can reinstall dependencies to fix this issue.`)
|
|
513
513
|
}
|
|
514
514
|
return binding
|
|
515
515
|
} catch (e) {
|
package/models/README.md
CHANGED
|
@@ -1,66 +1,42 @@
|
|
|
1
1
|
# Models
|
|
2
2
|
|
|
3
3
|
This directory holds the third-party model files decibri bundles, together with
|
|
4
|
-
the
|
|
5
|
-
|
|
6
|
-
|
|
4
|
+
the tensor interface each one exposes. The license notice every model requires
|
|
5
|
+
is in `THIRD-PARTY-NOTICES.md` beside this file, and ships alongside the weights
|
|
6
|
+
it covers inside the published npm and PyPI packages.
|
|
7
7
|
|
|
8
8
|
## silero_vad.onnx
|
|
9
9
|
|
|
10
|
-
- **
|
|
11
|
-
- **
|
|
12
|
-
- **
|
|
13
|
-
- **Size:** ~2.2 MB
|
|
10
|
+
- **Model:** Silero VAD v6.2
|
|
11
|
+
- **Source:** https://github.com/snakers4/silero-vad (release `v6.2`)
|
|
12
|
+
- **Size:** ~2.2 MB (2,327,524 bytes)
|
|
14
13
|
- **Purpose:** Voice Activity Detection. Determines whether an audio frame contains human speech
|
|
15
14
|
|
|
16
15
|
### Input/Output Specification
|
|
17
16
|
|
|
17
|
+
Verified from the model file (ONNX IR version 8, opset 16, exported with spox).
|
|
18
|
+
All tensors are fp32 except `sr`, which is i64. The batch, sample-count and
|
|
19
|
+
state dimensions are declared dynamic; the shapes below are what the model takes
|
|
20
|
+
and produces at batch 1.
|
|
21
|
+
|
|
18
22
|
**Inputs:**
|
|
19
|
-
- `input`: f32[batch, window_size]: audio samples (512 at
|
|
20
|
-
- `
|
|
21
|
-
- `
|
|
22
|
-
- `c`: f32[2, batch, 64]: LSTM cell state
|
|
23
|
+
- `input`: f32[batch, context_size + window_size]: audio samples. Each call is fed the previous window's last `context_size` samples followed by the current window (64 + 512 = 576 at 16 kHz, 32 + 256 = 288 at 8 kHz), where `context_size` is `window_size / 8`.
|
|
24
|
+
- `state`: f32[2, batch, 128]: combined LSTM hidden and cell state
|
|
25
|
+
- `sr`: i64 scalar: sample rate
|
|
23
26
|
|
|
24
27
|
**Outputs:**
|
|
25
28
|
- `output`: f32[batch, 1]: speech probability (0.0 to 1.0)
|
|
26
|
-
- `
|
|
27
|
-
- `cn`: f32[2, batch, 64]: updated cell state
|
|
28
|
-
|
|
29
|
-
Note: Exact tensor names may differ. Verified at build time from the model file.
|
|
30
|
-
|
|
31
|
-
### License Notice
|
|
32
|
-
|
|
33
|
-
This model is a third-party artifact, not proprietary to decibri. It is
|
|
34
|
-
distributed under the MIT License, reproduced in full below.
|
|
35
|
-
|
|
36
|
-
```text
|
|
37
|
-
MIT License
|
|
29
|
+
- `stateN`: f32[2, batch, 128]: updated state
|
|
38
30
|
|
|
39
|
-
|
|
40
|
-
|
|
41
|
-
|
|
42
|
-
of
|
|
43
|
-
|
|
44
|
-
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
|
|
45
|
-
copies of the Software, and to permit persons to whom the Software is
|
|
46
|
-
furnished to do so, subject to the following conditions:
|
|
47
|
-
|
|
48
|
-
The above copyright notice and this permission notice shall be included in all
|
|
49
|
-
copies or substantial portions of the Software.
|
|
50
|
-
|
|
51
|
-
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
|
|
52
|
-
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
|
|
53
|
-
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
|
|
54
|
-
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
|
|
55
|
-
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
|
|
56
|
-
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
|
|
57
|
-
SOFTWARE.
|
|
58
|
-
```
|
|
31
|
+
Streaming contract: the model carries no state of its own between calls. Initialize
|
|
32
|
+
`state` to zeros and the audio context to zeros at the start of a stream, feed each
|
|
33
|
+
call's `stateN` back as the next call's `state`, and carry the last `context_size`
|
|
34
|
+
samples of each window forward as the next call's context. Feeding a bare window
|
|
35
|
+
with no context prepended scores even loud speech at the non-speech floor.
|
|
59
36
|
|
|
60
37
|
## fastenhancer_t.onnx
|
|
61
38
|
|
|
62
|
-
- **
|
|
63
|
-
- **License:** MIT
|
|
39
|
+
- **Model:** FastEnhancer-T (tiny tier), VoiceBank-DEMAND checkpoint, waveform variant
|
|
64
40
|
- **Source:** https://github.com/aask1357/fastenhancer (release `onnx-vd-v1.0.0`)
|
|
65
41
|
- **Size:** ~122 KB (125,036 bytes)
|
|
66
42
|
- **Purpose:** Single-channel speech enhancement (denoise). Maps a window of noisy speech samples to a hop of cleaned speech samples, frame by frame, for streaming use.
|
|
@@ -89,61 +65,3 @@ receives audio samples with no spectral processing of its own. The three cache
|
|
|
89
65
|
tensors hold the model's overlap-add and recurrent state. Initialize them to
|
|
90
66
|
zeros at the start of a stream and feed each call's `cache_out` values back as
|
|
91
67
|
the next call's `cache_in` values.
|
|
92
|
-
|
|
93
|
-
### License Notice
|
|
94
|
-
|
|
95
|
-
This model is a third-party artifact, not proprietary to decibri. The model
|
|
96
|
-
code and weights are distributed under the MIT License, reproduced in full
|
|
97
|
-
below.
|
|
98
|
-
|
|
99
|
-
```text
|
|
100
|
-
MIT License
|
|
101
|
-
|
|
102
|
-
Copyright (c) 2025 AHN Sung Hwan
|
|
103
|
-
|
|
104
|
-
Permission is hereby granted, free of charge, to any person obtaining a copy
|
|
105
|
-
of this software and associated documentation files (the "Software"), to deal
|
|
106
|
-
in the Software without restriction, including without limitation the rights
|
|
107
|
-
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
|
|
108
|
-
copies of the Software, and to permit persons to whom the Software is
|
|
109
|
-
furnished to do so, subject to the following conditions:
|
|
110
|
-
|
|
111
|
-
The above copyright notice and this permission notice shall be included in all
|
|
112
|
-
copies or substantial portions of the Software.
|
|
113
|
-
|
|
114
|
-
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
|
|
115
|
-
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
|
|
116
|
-
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
|
|
117
|
-
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
|
|
118
|
-
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
|
|
119
|
-
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
|
|
120
|
-
SOFTWARE.
|
|
121
|
-
```
|
|
122
|
-
|
|
123
|
-
### Training-Data Attribution
|
|
124
|
-
|
|
125
|
-
The bundled checkpoint is trained on the VoiceBank-DEMAND noisy speech dataset,
|
|
126
|
-
which pairs clean speech from the CSTR VCTK Corpus with noise from the DEMAND
|
|
127
|
-
database. Each source dataset requires attribution:
|
|
128
|
-
|
|
129
|
-
- **VoiceBank-DEMAND.** Valentini-Botinhao, Cassia. (2017). Noisy speech
|
|
130
|
-
database for training speech enhancement algorithms and TTS models, 2016
|
|
131
|
-
[sound]. University of Edinburgh, School of Informatics, Centre for Speech
|
|
132
|
-
Technology Research (CSTR). Licensed under Creative Commons Attribution 4.0
|
|
133
|
-
International (CC BY 4.0), https://creativecommons.org/licenses/by/4.0/.
|
|
134
|
-
https://doi.org/10.7488/ds/2117
|
|
135
|
-
|
|
136
|
-
- **CSTR VCTK Corpus (version 0.92).** Yamagishi, Junichi; Veaux, Christophe;
|
|
137
|
-
MacDonald, Kirsten. (2019). CSTR VCTK Corpus: English Multi-speaker Corpus
|
|
138
|
-
for CSTR Voice Cloning Toolkit (version 0.92) [sound]. University of
|
|
139
|
-
Edinburgh, Centre for Speech Technology Research (CSTR). Licensed under the
|
|
140
|
-
Open Data Commons Attribution License (ODC-By) v1.0,
|
|
141
|
-
https://opendatacommons.org/licenses/by/1-0/.
|
|
142
|
-
https://doi.org/10.7488/ds/2645
|
|
143
|
-
|
|
144
|
-
- **DEMAND.** Thiemann, Joachim; Ito, Nobutaka; Vincent, Emmanuel. (2013).
|
|
145
|
-
DEMAND: a collection of multi-channel recordings of acoustic noise in
|
|
146
|
-
diverse environments. Licensed under Creative Commons
|
|
147
|
-
Attribution-ShareAlike 3.0 Unported (CC BY-SA 3.0),
|
|
148
|
-
https://creativecommons.org/licenses/by-sa/3.0/.
|
|
149
|
-
https://doi.org/10.5281/zenodo.1227121
|
|
@@ -0,0 +1,110 @@
|
|
|
1
|
+
# Third-Party Notices
|
|
2
|
+
|
|
3
|
+
This directory redistributes the model files listed below, together with the
|
|
4
|
+
license notices they require. The notices ship inside the published npm and
|
|
5
|
+
PyPI packages alongside the model weights they cover, so the attribution
|
|
6
|
+
travels with the files. `README.md` beside this file documents each model's
|
|
7
|
+
tensor interface.
|
|
8
|
+
|
|
9
|
+
## Silero VAD
|
|
10
|
+
|
|
11
|
+
- **Name:** Silero VAD
|
|
12
|
+
- **Version:** v6.2
|
|
13
|
+
- **License:** MIT
|
|
14
|
+
- **Source:** https://github.com/snakers4/silero-vad (release `v6.2`)
|
|
15
|
+
- **Files covered:** `silero_vad.onnx`
|
|
16
|
+
|
|
17
|
+
### License Notice
|
|
18
|
+
|
|
19
|
+
This model is a third-party artifact, not proprietary to decibri. It is
|
|
20
|
+
distributed under the MIT License, reproduced in full below.
|
|
21
|
+
|
|
22
|
+
```text
|
|
23
|
+
MIT License
|
|
24
|
+
|
|
25
|
+
Copyright (c) 2020-present Silero Team
|
|
26
|
+
|
|
27
|
+
Permission is hereby granted, free of charge, to any person obtaining a copy
|
|
28
|
+
of this software and associated documentation files (the "Software"), to deal
|
|
29
|
+
in the Software without restriction, including without limitation the rights
|
|
30
|
+
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
|
|
31
|
+
copies of the Software, and to permit persons to whom the Software is
|
|
32
|
+
furnished to do so, subject to the following conditions:
|
|
33
|
+
|
|
34
|
+
The above copyright notice and this permission notice shall be included in all
|
|
35
|
+
copies or substantial portions of the Software.
|
|
36
|
+
|
|
37
|
+
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
|
|
38
|
+
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
|
|
39
|
+
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
|
|
40
|
+
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
|
|
41
|
+
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
|
|
42
|
+
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
|
|
43
|
+
SOFTWARE.
|
|
44
|
+
```
|
|
45
|
+
|
|
46
|
+
## FastEnhancer
|
|
47
|
+
|
|
48
|
+
- **Name:** FastEnhancer-T (tiny tier), VoiceBank-DEMAND checkpoint, waveform variant
|
|
49
|
+
- **Version:** `onnx-vd-v1.0.0`
|
|
50
|
+
- **License:** MIT
|
|
51
|
+
- **Source:** https://github.com/aask1357/fastenhancer (release `onnx-vd-v1.0.0`)
|
|
52
|
+
- **Files covered:** `fastenhancer_t.onnx`
|
|
53
|
+
|
|
54
|
+
### License Notice
|
|
55
|
+
|
|
56
|
+
This model is a third-party artifact, not proprietary to decibri. The model
|
|
57
|
+
code and weights are distributed under the MIT License, reproduced in full
|
|
58
|
+
below.
|
|
59
|
+
|
|
60
|
+
```text
|
|
61
|
+
MIT License
|
|
62
|
+
|
|
63
|
+
Copyright (c) 2025 AHN Sung Hwan
|
|
64
|
+
|
|
65
|
+
Permission is hereby granted, free of charge, to any person obtaining a copy
|
|
66
|
+
of this software and associated documentation files (the "Software"), to deal
|
|
67
|
+
in the Software without restriction, including without limitation the rights
|
|
68
|
+
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
|
|
69
|
+
copies of the Software, and to permit persons to whom the Software is
|
|
70
|
+
furnished to do so, subject to the following conditions:
|
|
71
|
+
|
|
72
|
+
The above copyright notice and this permission notice shall be included in all
|
|
73
|
+
copies or substantial portions of the Software.
|
|
74
|
+
|
|
75
|
+
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
|
|
76
|
+
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
|
|
77
|
+
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
|
|
78
|
+
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
|
|
79
|
+
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
|
|
80
|
+
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
|
|
81
|
+
SOFTWARE.
|
|
82
|
+
```
|
|
83
|
+
|
|
84
|
+
### Training-Data Attribution
|
|
85
|
+
|
|
86
|
+
The bundled checkpoint is trained on the VoiceBank-DEMAND noisy speech dataset,
|
|
87
|
+
which pairs clean speech from the CSTR VCTK Corpus with noise from the DEMAND
|
|
88
|
+
database. Each source dataset requires attribution:
|
|
89
|
+
|
|
90
|
+
- **VoiceBank-DEMAND.** Valentini-Botinhao, Cassia. (2017). Noisy speech
|
|
91
|
+
database for training speech enhancement algorithms and TTS models, 2016
|
|
92
|
+
[sound]. University of Edinburgh, School of Informatics, Centre for Speech
|
|
93
|
+
Technology Research (CSTR). Licensed under Creative Commons Attribution 4.0
|
|
94
|
+
International (CC BY 4.0), https://creativecommons.org/licenses/by/4.0/.
|
|
95
|
+
https://doi.org/10.7488/ds/2117
|
|
96
|
+
|
|
97
|
+
- **CSTR VCTK Corpus (version 0.92).** Yamagishi, Junichi; Veaux, Christophe;
|
|
98
|
+
MacDonald, Kirsten. (2019). CSTR VCTK Corpus: English Multi-speaker Corpus
|
|
99
|
+
for CSTR Voice Cloning Toolkit (version 0.92) [sound]. University of
|
|
100
|
+
Edinburgh, Centre for Speech Technology Research (CSTR). Licensed under the
|
|
101
|
+
Open Data Commons Attribution License (ODC-By) v1.0,
|
|
102
|
+
https://opendatacommons.org/licenses/by/1-0/.
|
|
103
|
+
https://doi.org/10.7488/ds/2645
|
|
104
|
+
|
|
105
|
+
- **DEMAND.** Thiemann, Joachim; Ito, Nobutaka; Vincent, Emmanuel. (2013).
|
|
106
|
+
DEMAND: a collection of multi-channel recordings of acoustic noise in
|
|
107
|
+
diverse environments. Licensed under Creative Commons
|
|
108
|
+
Attribution-ShareAlike 3.0 Unported (CC BY-SA 3.0),
|
|
109
|
+
https://creativecommons.org/licenses/by-sa/3.0/.
|
|
110
|
+
https://doi.org/10.5281/zenodo.1227121
|
package/package.json
CHANGED
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
{
|
|
2
2
|
"name": "decibri",
|
|
3
|
-
"version": "5.
|
|
3
|
+
"version": "5.5.0",
|
|
4
4
|
"description": "Cross-platform audio capture, playback, and processing for Node.js and browsers",
|
|
5
5
|
"main": "src/decibri.js",
|
|
6
6
|
"types": "src/decibri.d.ts",
|
|
@@ -72,10 +72,10 @@
|
|
|
72
72
|
"MIGRATION.md"
|
|
73
73
|
],
|
|
74
74
|
"optionalDependencies": {
|
|
75
|
-
"@decibri/decibri-win32-x64-msvc": "5.
|
|
76
|
-
"@decibri/decibri-darwin-arm64": "5.
|
|
77
|
-
"@decibri/decibri-linux-x64-gnu": "5.
|
|
78
|
-
"@decibri/decibri-linux-arm64-gnu": "5.
|
|
75
|
+
"@decibri/decibri-win32-x64-msvc": "5.5.0",
|
|
76
|
+
"@decibri/decibri-darwin-arm64": "5.5.0",
|
|
77
|
+
"@decibri/decibri-linux-x64-gnu": "5.5.0",
|
|
78
|
+
"@decibri/decibri-linux-arm64-gnu": "5.5.0"
|
|
79
79
|
},
|
|
80
80
|
"devDependencies": {
|
|
81
81
|
"@napi-rs/cli": "^3.7.0"
|
|
@@ -6,7 +6,7 @@ const { WORKLET_SOURCE } = require('./worklet-inline.js');
|
|
|
6
6
|
// Browser build version. Keep in sync with package.json on each release; the
|
|
7
7
|
// browser bundle cannot read package.json at runtime the way the Node wrapper
|
|
8
8
|
// does, so this is a maintained constant.
|
|
9
|
-
const VERSION = '5.
|
|
9
|
+
const VERSION = '5.5.0';
|
|
10
10
|
|
|
11
11
|
/**
|
|
12
12
|
* Browser microphone capture.
|
|
@@ -105,6 +105,19 @@ class Microphone extends Emitter {
|
|
|
105
105
|
if (this._sampleRate < 1000 || this._sampleRate > 384000) {
|
|
106
106
|
throw new RangeError('sample rate must be between 1000 and 384000');
|
|
107
107
|
}
|
|
108
|
+
// Mono only: the worklet reads a single channel. The classes and the
|
|
109
|
+
// messages are the node entry's, so the same value is rejected the same
|
|
110
|
+
// way in both runtimes.
|
|
111
|
+
if (this._channels < 1) {
|
|
112
|
+
throw new RangeError('channels must be at least 1');
|
|
113
|
+
}
|
|
114
|
+
if (this._channels > 1) {
|
|
115
|
+
throw new RangeError('multichannel capture is not supported; channels must be 1 (mono)');
|
|
116
|
+
}
|
|
117
|
+
// The Web Audio specification's floor, as on the browser Speaker: an
|
|
118
|
+
// implementation is required to support up to 32 channels and says nothing
|
|
119
|
+
// above that. Deliberately unlike the native surface, which bounds channels
|
|
120
|
+
// below only.
|
|
108
121
|
if (this._channels < 1 || this._channels > 32) {
|
|
109
122
|
throw new TypeError(`channels must be between 1 and 32, got ${this._channels}`);
|
|
110
123
|
}
|
|
@@ -104,6 +104,12 @@ class Speaker {
|
|
|
104
104
|
if (this._sampleRate < 1000 || this._sampleRate > 384000) {
|
|
105
105
|
throw new TypeError(`sample rate must be between 1000 and 384000, got ${this._sampleRate}`);
|
|
106
106
|
}
|
|
107
|
+
// The 32 is the Web Audio specification's floor: an implementation is
|
|
108
|
+
// required to support up to 32 channels and the specification says nothing
|
|
109
|
+
// above that, so 32 is what a browser can be relied on to accept. The
|
|
110
|
+
// native surface bounds output channels below only and leaves the maximum
|
|
111
|
+
// to the device; the two differ deliberately, because they are answering to
|
|
112
|
+
// different things.
|
|
107
113
|
if (this._channels < 1 || this._channels > 32) {
|
|
108
114
|
throw new TypeError(`channels must be between 1 and 32, got ${this._channels}`);
|
|
109
115
|
}
|
package/src/browser/index.d.ts
CHANGED
|
@@ -47,9 +47,12 @@ export interface MicrophoneOptions {
|
|
|
47
47
|
sampleRate?: number;
|
|
48
48
|
|
|
49
49
|
/**
|
|
50
|
-
* Number of input channels.
|
|
50
|
+
* Number of input channels. Mono only: the only accepted value is `1`, and a
|
|
51
|
+
* value greater than `1` throws a `RangeError` (multichannel capture is not
|
|
52
|
+
* supported) rather than being silently downmixed. The option is kept for
|
|
53
|
+
* forward compatibility: a future release may accept a value greater than `1`
|
|
54
|
+
* by delivering true interleaved multichannel.
|
|
51
55
|
* @default 1
|
|
52
|
-
* @range 1–32
|
|
53
56
|
*/
|
|
54
57
|
channels?: number;
|
|
55
58
|
|
package/src/decibri-output.js
CHANGED
|
@@ -67,9 +67,13 @@ class Speaker extends Writable {
|
|
|
67
67
|
throw new RangeError('sample rate must be between 1000 and 384000');
|
|
68
68
|
}
|
|
69
69
|
|
|
70
|
+
// Bounded below only. How many output channels can be carried is the
|
|
71
|
+
// device's answer, so any count above zero is passed through and a device
|
|
72
|
+
// that refuses it throws a DecibriError with code
|
|
73
|
+
// 'SPEAKER_CHANNELS_UNSUPPORTED' naming the count the device reports.
|
|
70
74
|
const channels = options.channels ?? 1;
|
|
71
|
-
if (channels < 1
|
|
72
|
-
throw new RangeError('channels must be
|
|
75
|
+
if (channels < 1) {
|
|
76
|
+
throw new RangeError('channels must be at least 1');
|
|
73
77
|
}
|
|
74
78
|
|
|
75
79
|
const dtype = options.dtype ?? 'int16';
|
package/src/decibri.d.ts
CHANGED
|
@@ -42,7 +42,7 @@ export interface VersionInfo {
|
|
|
42
42
|
export interface VadOptions {
|
|
43
43
|
/**
|
|
44
44
|
* Which detector to run.
|
|
45
|
-
* - `'silero'`: Silero VAD
|
|
45
|
+
* - `'silero'`: Silero VAD v6.2 ML model (more accurate, ~1ms inference)
|
|
46
46
|
* - `'energy'`: RMS energy threshold (lightweight, no model)
|
|
47
47
|
*/
|
|
48
48
|
model: 'silero' | 'energy';
|
|
@@ -105,6 +105,32 @@ export interface AecOptions {
|
|
|
105
105
|
* @range 1000 to 384000
|
|
106
106
|
*/
|
|
107
107
|
referenceSampleRate?: number;
|
|
108
|
+
|
|
109
|
+
/**
|
|
110
|
+
* Number of channels in the far-end reference pushed through
|
|
111
|
+
* `pushAecReference`, frame-interleaved. When it names a count above 1,
|
|
112
|
+
* decibri averages each frame to one mono sample before the canceller sees
|
|
113
|
+
* it: a multichannel reference pushed without declaring the count cancels
|
|
114
|
+
* nothing and reports no error, so the collapse is decibri's rather than
|
|
115
|
+
* the caller's.
|
|
116
|
+
*
|
|
117
|
+
* The declared count must match the buffer actually pushed. The reference
|
|
118
|
+
* arrives as flat PCM whose true channel count is not recoverable from its
|
|
119
|
+
* length, so a mismatch is not detected and raises no error: the frames
|
|
120
|
+
* are misread, nothing is cancelled, and the observable signature is
|
|
121
|
+
* `aecMetrics().delaySamples` staying `null` while the canceller reports
|
|
122
|
+
* no fault.
|
|
123
|
+
*
|
|
124
|
+
* The canceller itself reads one mono reference. Against playback through
|
|
125
|
+
* more than one loudspeaker that is a cancellation ceiling: the echo
|
|
126
|
+
* reaching the microphone is the sum of different room responses driven by
|
|
127
|
+
* different signals, and a single-reference canceller models one response
|
|
128
|
+
* applied to their average, so a placement where those paths differ leaves
|
|
129
|
+
* a residual that no amount of adaptation removes.
|
|
130
|
+
* @default 1 (mono)
|
|
131
|
+
* @range at least 1; no upper bound
|
|
132
|
+
*/
|
|
133
|
+
referenceChannels?: number;
|
|
108
134
|
}
|
|
109
135
|
|
|
110
136
|
/**
|
|
@@ -211,7 +237,7 @@ export interface MicrophoneOptions extends ReadableOptions {
|
|
|
211
237
|
/**
|
|
212
238
|
* Voice activity detection. One of:
|
|
213
239
|
* - `false`: disabled (default)
|
|
214
|
-
* - `'silero'`: Silero VAD
|
|
240
|
+
* - `'silero'`: Silero VAD v6.2 ML model (more accurate, ~1ms inference)
|
|
215
241
|
* - `'energy'`: RMS energy threshold (lightweight)
|
|
216
242
|
* - a `VadOptions` config object `{ model, threshold?, holdoffMs? }` to tune
|
|
217
243
|
* the threshold and holdoff for the chosen model
|
|
@@ -371,8 +397,14 @@ export declare class Microphone extends Readable {
|
|
|
371
397
|
* Queue far-end reference audio for the echo canceller: the audio being
|
|
372
398
|
* played out, pushed as it is played, in played order. Accepts the same
|
|
373
399
|
* input shapes `Speaker.write` accepts (a `Buffer`, any TypedArray, or a
|
|
374
|
-
* `DataView` of PCM bytes in this microphone's `dtype`),
|
|
375
|
-
*
|
|
400
|
+
* `DataView` of PCM bytes in this microphone's `dtype`), at the declared
|
|
401
|
+
* `referenceSampleRate` (the capture rate when unset), interleaved at the
|
|
402
|
+
* declared `referenceChannels` (mono when unset). With `referenceChannels`
|
|
403
|
+
* above 1, each frame is averaged to one mono sample before the canceller
|
|
404
|
+
* sees it. The declared count must match this buffer's actual
|
|
405
|
+
* interleaving: a mismatch is not detected and raises no error, and shows
|
|
406
|
+
* up only as `aecMetrics().delaySamples` staying `null` with no fault
|
|
407
|
+
* reported.
|
|
376
408
|
*
|
|
377
409
|
* Never blocks and never throws on a full queue: samples that do not fit
|
|
378
410
|
* are discarded and counted by `aecMetrics().referenceDropped`. Silence
|
|
@@ -810,9 +842,11 @@ export interface SpeakerOptions extends WritableOptions {
|
|
|
810
842
|
sampleRate?: number;
|
|
811
843
|
|
|
812
844
|
/**
|
|
813
|
-
* Number of output channels.
|
|
845
|
+
* Number of output channels. The maximum is the device's: a count the device
|
|
846
|
+
* cannot serve throws a `DecibriError` with code
|
|
847
|
+
* `'SPEAKER_CHANNELS_UNSUPPORTED'` naming the count the device reports.
|
|
814
848
|
* @default 1
|
|
815
|
-
* @range 1
|
|
849
|
+
* @range 1 or more
|
|
816
850
|
*/
|
|
817
851
|
channels?: number;
|
|
818
852
|
|
package/src/decibri.js
CHANGED
|
@@ -151,7 +151,7 @@ class Microphone extends Readable {
|
|
|
151
151
|
// additive. The `channels` option is kept for that forward compatibility.
|
|
152
152
|
const channels = options.channels ?? 1;
|
|
153
153
|
if (channels < 1) {
|
|
154
|
-
throw new RangeError('channels must be
|
|
154
|
+
throw new RangeError('channels must be at least 1');
|
|
155
155
|
}
|
|
156
156
|
if (channels > 1) {
|
|
157
157
|
throw new RangeError('multichannel capture is not supported; channels must be 1 (mono)');
|
|
@@ -345,8 +345,8 @@ class Microphone extends Readable {
|
|
|
345
345
|
// ── Validate AEC ─────────────────────────────────────────────────────────
|
|
346
346
|
|
|
347
347
|
// Echo cancellation: the 'tau' shorthand names the model, or an
|
|
348
|
-
// { model, tailMs, suppression, referenceSampleRate }
|
|
349
|
-
// absence leaves it off. The model name is deliberately NOT checked
|
|
348
|
+
// { model, tailMs, suppression, referenceSampleRate, referenceChannels }
|
|
349
|
+
// object tunes it; absence leaves it off. The model name is deliberately NOT checked
|
|
350
350
|
// against a list here: the canceller owns the accepted set, so the native
|
|
351
351
|
// layer parses it (AecModel::from_str) and an unknown name is rejected by
|
|
352
352
|
// the native constructor with the canceller's own message (a DecibriError
|
|
@@ -360,11 +360,12 @@ class Microphone extends Readable {
|
|
|
360
360
|
let aecTailMs;
|
|
361
361
|
let aecSuppression;
|
|
362
362
|
let aecReferenceSampleRate;
|
|
363
|
+
let aecReferenceChannels;
|
|
363
364
|
if (aec !== undefined) {
|
|
364
365
|
if (typeof aec === 'string') {
|
|
365
366
|
aecModel = aec;
|
|
366
367
|
} else if (aec !== null && typeof aec === 'object' && !Array.isArray(aec)) {
|
|
367
|
-
const { model, tailMs, suppression, referenceSampleRate } = aec;
|
|
368
|
+
const { model, tailMs, suppression, referenceSampleRate, referenceChannels } = aec;
|
|
368
369
|
if (typeof model !== 'string') {
|
|
369
370
|
throw new TypeError(
|
|
370
371
|
`Invalid aec model: ${JSON.stringify(model)}. Expected a model name string such as 'tau'.`
|
|
@@ -397,9 +398,18 @@ class Microphone extends Readable {
|
|
|
397
398
|
}
|
|
398
399
|
aecReferenceSampleRate = referenceSampleRate;
|
|
399
400
|
}
|
|
401
|
+
if (referenceChannels !== undefined) {
|
|
402
|
+
if (typeof referenceChannels !== 'number' || Number.isNaN(referenceChannels)) {
|
|
403
|
+
throw new TypeError('aec referenceChannels must be a number');
|
|
404
|
+
}
|
|
405
|
+
if (referenceChannels < 1) {
|
|
406
|
+
throw new RangeError('aec referenceChannels must be at least 1');
|
|
407
|
+
}
|
|
408
|
+
aecReferenceChannels = referenceChannels;
|
|
409
|
+
}
|
|
400
410
|
} else {
|
|
401
411
|
throw new TypeError(
|
|
402
|
-
`Invalid aec value: ${JSON.stringify(aec)}. Expected a model name such as 'tau', or a config object { model, tailMs, suppression, referenceSampleRate }.`
|
|
412
|
+
`Invalid aec value: ${JSON.stringify(aec)}. Expected a model name such as 'tau', or a config object { model, tailMs, suppression, referenceSampleRate, referenceChannels }.`
|
|
403
413
|
);
|
|
404
414
|
}
|
|
405
415
|
}
|
|
@@ -443,6 +453,7 @@ class Microphone extends Readable {
|
|
|
443
453
|
aecTailMs,
|
|
444
454
|
aecSuppression,
|
|
445
455
|
aecReferenceSampleRate,
|
|
456
|
+
aecReferenceChannels,
|
|
446
457
|
},
|
|
447
458
|
};
|
|
448
459
|
}
|
|
@@ -605,8 +616,15 @@ class Microphone extends Readable {
|
|
|
605
616
|
* Queue far-end reference audio for the echo canceller: the audio being
|
|
606
617
|
* played out, pushed as it is played, in played order. Accepts the same
|
|
607
618
|
* input shapes `Speaker.write` accepts (a `Buffer`, any TypedArray, or a
|
|
608
|
-
* `DataView` of PCM bytes in this microphone's `dtype`),
|
|
609
|
-
*
|
|
619
|
+
* `DataView` of PCM bytes in this microphone's `dtype`), at the declared
|
|
620
|
+
* `referenceSampleRate` (the capture rate when unset), interleaved at the
|
|
621
|
+
* declared `referenceChannels` (mono when unset). With `referenceChannels`
|
|
622
|
+
* above 1, each frame is averaged to one mono sample before the canceller
|
|
623
|
+
* sees it; a multichannel reference pushed without declaring the count
|
|
624
|
+
* cancels nothing and reports no error. The declared count must match this
|
|
625
|
+
* buffer's actual interleaving: a mismatch is not detected and raises no
|
|
626
|
+
* error, and shows up only as `aecMetrics().delaySamples` staying `null`
|
|
627
|
+
* with no fault reported.
|
|
610
628
|
*
|
|
611
629
|
* Never blocks and never throws on a full queue: samples that do not fit
|
|
612
630
|
* are discarded and counted by `aecMetrics().referenceDropped`, and the
|
package/src/errors.js
CHANGED
|
@@ -75,7 +75,7 @@ class OrtPathError extends OrtError {
|
|
|
75
75
|
// the core ever sees it).
|
|
76
76
|
const RANGE_PREFIXES = [
|
|
77
77
|
'sample rate must be between',
|
|
78
|
-
'channels must be
|
|
78
|
+
'channels must be at least',
|
|
79
79
|
'multichannel capture is not supported',
|
|
80
80
|
'frames per buffer must be between',
|
|
81
81
|
'agc target level must be between',
|
|
@@ -83,6 +83,7 @@ const RANGE_PREFIXES = [
|
|
|
83
83
|
'flac compression level must be between',
|
|
84
84
|
'aec tailMs must be between',
|
|
85
85
|
'aec referenceSampleRate must be between',
|
|
86
|
+
'aec referenceChannels must be at',
|
|
86
87
|
'Silero VAD only supports',
|
|
87
88
|
'VAD threshold must be between',
|
|
88
89
|
'echo cancellation only supports',
|
|
@@ -125,6 +126,7 @@ const BASE_CODES = [
|
|
|
125
126
|
['audio stream is already running', 'ALREADY_RUNNING'],
|
|
126
127
|
['Failed to open audio stream', 'STREAM_OPEN_FAILED'],
|
|
127
128
|
['Failed to start audio stream', 'STREAM_START_FAILED'],
|
|
129
|
+
['the output device does not support', 'SPEAKER_CHANNELS_UNSUPPORTED'],
|
|
128
130
|
['Microphone permission denied.', 'PERMISSION_DENIED'],
|
|
129
131
|
['Microphone stream is closed', 'MICROPHONE_STREAM_CLOSED'],
|
|
130
132
|
['Speaker stream is closed', 'SPEAKER_STREAM_CLOSED'],
|