Shubhra/ajs 37 refactor tts with streams #402
Conversation
|
| private input = new IdentityTransform<AudioFrame | typeof VADStream.FLUSH_SENTINEL>(); | ||
| private output = new IdentityTransform<VADEvent>(); | ||
| private metricsStream: ReadableStream<VADEvent>; | ||
| private outputReader: ReadableStreamDefaultReader<VADEvent>; | ||
| private inputWriter: WritableStreamDefaultWriter<AudioFrame | typeof VADStream.FLUSH_SENTINEL>; |
There was a problem hiding this comment.
Making access modifiers more restrictive . Forgot to do it in #390
| if (!this.inputClosed) { | ||
| this.inputWriter.close(); | ||
| } |
There was a problem hiding this comment.
forgot to do in #390
| SynthesizedAudio | typeof SynthesizeStream.END_OF_STREAM | ||
| >; | ||
| private logger = log(); | ||
| private inputClosed = false; |
There was a problem hiding this comment.
This is duplicative. The Readable/WritableStreamDefaultWriter internally tracks whether it's closed but doesn't expose it. The only way to know is when you try to write to or close an already-closed writer, it throws an error.
The other option would be to
try {
this.inputWriter.write(SynthesizeStream.FLUSH_SENTINEL);
} catch (error) {
throw new Error('Input is closed');
}everywhere, which doesn't seem much better. Let me know what you thoughts are @lukasIO @toubatbrian
Similar to #390