speechflow 2.0.0 → 2.0.1

package/speechflow-cli/src/speechflow-node-t2a-amazon.ts

@@ -9,6 +9,7 @@ import Stream from "node:stream"
 
 /*  external dependencies  */
 import { getStreamAsBuffer } from "get-stream"
+import { Duration }          from "luxon"
 import SpeexResampler        from "speex-resampler"
 import {
     PollyClient, SynthesizeSpeechCommand,
@@ -25,9 +26,9 @@ export default class SpeechFlowNodeT2AAmazon extends SpeechFlowNode {
     public static name = "t2a-amazon"
 
     /*  internal state  */
-    private client: PollyClient | null = null
-    private closing = false
+    private client:    PollyClient    | null = null
     private resampler: SpeexResampler | null = null
+    private closing                          = false
 
     /*  construct node  */
     constructor (id: string, cfg: { [ id: string ]: any }, opts: { [ id: string ]: any }, args: any[]) {
@@ -129,22 +130,43 @@ export default class SpeechFlowNodeT2AAmazon extends SpeechFlowNode {
                 }
                 if (Buffer.isBuffer(chunk.payload))
                     callback(new Error("invalid chunk payload type"))
-                else if (chunk.payload.length > 0) {
+                else if (chunk.payload === "")
+                    callback()
+                else {
+                    let processTimeout: ReturnType<typeof setTimeout> | null = setTimeout(() => {
+                        processTimeout = null
+                        callback(new Error("AWS Polly API timeout"))
+                    }, 60 * 1000)
+                    const clearProcessTimeout = () => {
+                        if (processTimeout !== null) {
+                            clearTimeout(processTimeout)
+                            processTimeout = null
+                        }
+                    }
                     self.log("debug", `send data (${chunk.payload.length} bytes): "${chunk.payload}"`)
                     textToSpeech(chunk.payload as string).then((buffer) => {
-                        if (self.closing)
-                            throw new Error("stream destroyed during processing")
+                        if (self.closing) {
+                            clearProcessTimeout()
+                            callback(new Error("stream destroyed during processing"))
+                            return
+                        }
+                        /*  calculate actual audio duration from PCM buffer size  */
+                        const durationMs = util.audioBufferDuration(buffer,
+                            self.config.audioSampleRate, self.config.audioBitDepth) * 1000
+
+                        /*  create new chunk with recalculated timestamps  */
                         const chunkNew = chunk.clone()
-                        chunkNew.type = "audio"
-                        chunkNew.payload = buffer
+                        chunkNew.type         = "audio"
+                        chunkNew.payload      = buffer
+                        chunkNew.timestampEnd = Duration.fromMillis(chunkNew.timestampStart.toMillis() + durationMs)
+                        clearProcessTimeout()
                         this.push(chunkNew)
                         callback()
                     }).catch((error: unknown) => {
-                        callback(util.ensureError(error, "failed to send to AWS Polly"))
+                        clearProcessTimeout()
+                        callback(util.ensureError(error, "AWS Polly processing failed"))
                     })
                 }
-                else
-                    callback()
             },
             final (callback) {
                 callback()

package/speechflow-cli/src/speechflow-node-t2a-elevenlabs.ts

@@ -24,8 +24,8 @@ export default class SpeechFlowNodeT2AElevenlabs extends SpeechFlowNode {
 
     /*  internal state  */
     private elevenlabs: ElevenLabs.ElevenLabsClient | null = null
-    private closing = false
-    private resampler: SpeexResampler | null = null
+    private resampler:  SpeexResampler              | null = null
+    private closing                                        = false
 
     /*  construct node  */
     constructor (id: string, cfg: { [ id: string ]: any }, opts: { [ id: string ]: any }, args: any[]) {
@@ -131,8 +131,8 @@ export default class SpeechFlowNodeT2AElevenlabs extends SpeechFlowNode {
             })
         }
 
-        /*  establish resampler from ElevenLabs's maximum 24Khz
-            output to our standard audio sample rate (48KHz)  */
+        /*  establish resampler from ElevenLabs's tier-dependent
+            output sample rate to our standard audio sample rate (48KHz)  */
         this.resampler = new SpeexResampler(1, maxSampleRate, this.config.audioSampleRate, 7)
 
         /*  create transform stream and connect it to the ElevenLabs API  */
@@ -147,6 +147,8 @@ export default class SpeechFlowNodeT2AElevenlabs extends SpeechFlowNode {
                     callback(new Error("stream already destroyed"))
                 else if (Buffer.isBuffer(chunk.payload))
                     callback(new Error("invalid chunk payload type"))
+                else if (chunk.payload === "")
+                    callback()
                 else {
                     let processTimeout: ReturnType<typeof setTimeout> | null = setTimeout(() => {
                         processTimeout = null

package/speechflow-cli/src/speechflow-node-t2a-google.ts

@@ -126,11 +126,8 @@ export default class SpeechFlowNodeT2AGoogle extends SpeechFlowNode {
                     callback(new Error("stream already destroyed"))
                 else if (Buffer.isBuffer(chunk.payload))
                     callback(new Error("invalid chunk payload type"))
-                else if (chunk.payload === "") {
-                    /*  pass through empty chunks  */
-                    this.push(chunk)
+                else if (chunk.payload === "")
                     callback()
-                }
                 else {
                     let processTimeout: ReturnType<typeof setTimeout> | null = setTimeout(() => {
                         processTimeout = null

package/speechflow-cli/src/speechflow-node-t2a-kokoro.ts

@@ -9,6 +9,7 @@ import Stream from "node:stream"
 
 /*  external dependencies  */
 import { KokoroTTS }  from "kokoro-js"
+import { Duration }   from "luxon"
 import SpeexResampler from "speex-resampler"
 
 /*  internal dependencies  */
@@ -21,9 +22,9 @@ export default class SpeechFlowNodeT2AKokoro extends SpeechFlowNode {
     public static name = "t2a-kokoro"
 
     /*  internal state  */
-    private kokoro: KokoroTTS | null = null
-    private closing = false
+    private kokoro:    KokoroTTS      | null = null
     private resampler: SpeexResampler | null = null
+    private closing                          = false
 
     /*  construct node  */
     constructor (id: string, cfg: { [ id: string ]: any }, opts: { [ id: string ]: any }, args: any[]) {
@@ -122,9 +123,7 @@ export default class SpeechFlowNodeT2AKokoro extends SpeechFlowNode {
             }
 
             /*  resample audio samples from PCM/I16/24Khz to PCM/I16/48KHz  */
-            const buffer2 = this.resampler!.processChunk(buffer1)
-
-            return buffer2
+            return this.resampler!.processChunk(buffer1)
         }
 
         /*  create transform stream and connect it to the Kokoro API  */
@@ -139,18 +138,42 @@ export default class SpeechFlowNodeT2AKokoro extends SpeechFlowNode {
                     callback(new Error("stream already destroyed"))
                 else if (Buffer.isBuffer(chunk.payload))
                     callback(new Error("invalid chunk payload type"))
+                else if (chunk.payload === "")
+                    callback()
                 else {
+                    let processTimeout: ReturnType<typeof setTimeout> | null = setTimeout(() => {
+                        processTimeout = null
+                        callback(new Error("Kokoro TTS timeout"))
+                    }, 60 * 1000)
+                    const clearProcessTimeout = () => {
+                        if (processTimeout !== null) {
+                            clearTimeout(processTimeout)
+                            processTimeout = null
+                        }
+                    }
                     text2speech(chunk.payload).then((buffer) => {
-                        if (self.closing)
-                            throw new Error("stream destroyed during processing")
+                        if (self.closing) {
+                            clearProcessTimeout()
+                            callback(new Error("stream destroyed during processing"))
+                            return
+                        }
                         self.log("info", `Kokoro: received audio (buffer length: ${buffer.byteLength})`)
+
+                        /*  calculate actual audio duration from PCM buffer size  */
+                        const durationMs = util.audioBufferDuration(buffer,
+                            self.config.audioSampleRate, self.config.audioBitDepth) * 1000
+
+                        /*  create new chunk with recalculated timestamps  */
                         const chunkNew = chunk.clone()
-                        chunkNew.type = "audio"
-                        chunkNew.payload = buffer
+                        chunkNew.type         = "audio"
+                        chunkNew.payload      = buffer
+                        chunkNew.timestampEnd = Duration.fromMillis(chunkNew.timestampStart.toMillis() + durationMs)
+                        clearProcessTimeout()
                         this.push(chunkNew)
                         callback()
                     }).catch((error: unknown) => {
-                        callback(util.ensureError(error))
+                        clearProcessTimeout()
+                        callback(util.ensureError(error, "Kokoro processing failed"))
                     })
                 }
             },

package/speechflow-cli/src/speechflow-node-t2a-openai.ts

@@ -103,11 +103,8 @@ export default class SpeechFlowNodeT2AOpenAI extends SpeechFlowNode {
                     callback(new Error("stream already destroyed"))
                 else if (Buffer.isBuffer(chunk.payload))
                     callback(new Error("invalid chunk payload type"))
-                else if (chunk.payload === "") {
-                    /*  pass through empty chunks  */
-                    this.push(chunk)
+                else if (chunk.payload === "")
                     callback()
-                }
                 else {
                     let processTimeout: ReturnType<typeof setTimeout> | null = setTimeout(() => {
                         processTimeout = null

package/speechflow-cli/src/speechflow-node-t2a-supertonic.ts

@@ -131,6 +131,7 @@ function chunkText (text: string, maxLen = 300): string[] {
 class SupertonicTextProcessor {
     private indexer: Record<number, number>
 
+    /*  construct text processor  */
     constructor (unicodeIndexerJsonPath: string) {
         /*  load and parse unicode indexer JSON  */
         try {
@@ -141,6 +142,7 @@ class SupertonicTextProcessor {
         }
     }
 
+    /*  preprocess text  */
     private preprocessText (text: string): string {
         /*  normalize text  */
         text = text.normalize("NFKD")
@@ -211,11 +213,13 @@ class SupertonicTextProcessor {
         return text
     }
 
+    /*  convert text to Unicode values  */
     private textToUnicodeValues (text: string): number[] {
         /*  convert text characters to unicode code points  */
         return Array.from(text).map((char) => char.charCodeAt(0))
     }
 
+    /*  process text list  */
     call (textList: string[]): { textIds: number[][], textMask: number[][][] } {
         /*  handle empty input  */
         if (textList.length === 0)
@@ -246,6 +250,7 @@ class SupertonicTextProcessor {
 class SupertonicTTS {
     public  sampleRate:          number
 
+    /*  internal TTS state  */
     private cfgs:                SupertonicConfig
     private textProcessor:       SupertonicTextProcessor
     private dpOrt:               ORT.InferenceSession
@@ -256,6 +261,7 @@ class SupertonicTTS {
     private chunkCompressFactor: number
     private latentDim:           number
 
+    /*  construct TTS engine  */
     constructor (
         cfgs:          SupertonicConfig,
         textProcessor: SupertonicTextProcessor,
@@ -279,6 +285,7 @@ class SupertonicTTS {
         this.latentDim           = cfgs.ttl.latent_dim
     }
 
+    /*  sample noisy latent vectors  */
     private sampleNoisyLatent (duration: number[]): { noisyLatent: number[][][], latentMask: number[][][] } {
         /*  calculate dimensions for latent space  */
         const wavLenMax  = Math.max(...duration) * this.sampleRate
@@ -294,7 +301,6 @@ class SupertonicTTS {
             for (let d = 0; d < latentDimExpanded; d++) {
                 const row: number[] = Array.from({ length: latentLen })
                 for (let t = 0; t < latentLen; t++) {
-
                     /*  Box-Muller transform for normal distribution  */
                     const eps = 1e-10
                     const u1 = Math.max(eps, Math.random())
@@ -317,6 +323,7 @@ class SupertonicTTS {
         return { noisyLatent, latentMask }
     }
 
+    /*  perform inference  */
     private async infer (textList: string[], style: SupertonicStyle, totalStep: number, speed: number): Promise<{ wav: number[], duration: number[] }> {
         /*  validate batch size matches style vectors  */
         if (textList.length !== style.ttl.dims[0])
@@ -392,6 +399,7 @@ class SupertonicTTS {
         return { wav, duration: predictedDurations }
     }
 
+    /*  synthesize speech from text  */
     async synthesize (text: string, style: SupertonicStyle, totalStep: number, speed: number, silenceDuration = 0.3): Promise<{ wav: number[], duration: number }> {
         /*  validate single speaker mode  */
         if (style.ttl.dims[0] !== 1)
@@ -420,6 +428,7 @@ class SupertonicTTS {
         return { wav: wavParts.flat(), duration: totalDuration }
     }
 
+    /*  release TTS engine resources  */
     async release (): Promise<void> {
         /*  release all ONNX inference sessions  */
         await Promise.all([
@@ -535,7 +544,7 @@ export default class SpeechFlowNodeT2ASupertonic extends SpeechFlowNode {
             "onnx/text_encoder.onnx",
             "onnx/unicode_indexer.json",
             "onnx/vector_estimator.onnx",
-            "onnx/vocoder.onnx",
+            "onnx/vocoder.onnx"
         ]
 
         /*  create asset directories  */
@@ -602,9 +611,8 @@ export default class SpeechFlowNodeT2ASupertonic extends SpeechFlowNode {
                 buffer1.writeInt16LE(sample * 0x7FFF, i * 2)
             }
 
-            /*  resample audio samples from 44.1kHz to 48kHz  */
-            const buffer2 = this.resampler!.processChunk(buffer1)
-            return buffer2
+            /*  resample audio samples from Supertonic sample rate to 48kHz  */
+            return this.resampler!.processChunk(buffer1)
         }
 
         /*  create transform stream and connect it to the Supertonic TTS  */
@@ -619,6 +627,8 @@ export default class SpeechFlowNodeT2ASupertonic extends SpeechFlowNode {
                     callback(new Error("stream already destroyed"))
                 else if (Buffer.isBuffer(chunk.payload))
                     callback(new Error("invalid chunk payload type"))
+                else if (chunk.payload === "")
+                    callback()
                 else {
                     let processTimeout: ReturnType<typeof setTimeout> | null = setTimeout(() => {
                         processTimeout = null
@@ -660,7 +670,6 @@ export default class SpeechFlowNodeT2ASupertonic extends SpeechFlowNode {
                         callback()
                     }
                     catch (error) {
-
                         /*  handle processing errors  */
                         clearProcessTimeout()
                         callback(util.ensureError(error, "Supertonic processing failed"))

package/speechflow-cli/src/speechflow-node-t2t-amazon.ts

@@ -5,7 +5,7 @@
 */
 
 /*  standard dependencies  */
-import Stream from "node:stream"
+import Stream                                    from "node:stream"
 
 /*  external dependencies  */
 import { TranslateClient, TranslateTextCommand } from "@aws-sdk/client-translate"
@@ -65,8 +65,6 @@ export default class SpeechFlowNodeT2TAmazon extends SpeechFlowNode {
                 secretAccessKey: this.params.secKey
             }
         })
-        if (this.client === null)
-            throw new Error("failed to establish Amazon Translate client")
 
         /*  provide text-to-text translation  */
         const maxRetries = 10

package/speechflow-cli/src/speechflow-node-t2t-deepl.ts

@@ -5,10 +5,10 @@
 */
 
 /*  standard dependencies  */
-import Stream from "node:stream"
+import Stream                              from "node:stream"
 
 /*  external dependencies  */
-import * as DeepL from "deepl-node"
+import * as DeepL                          from "deepl-node"
 
 /*  internal dependencies  */
 import SpeechFlowNode, { SpeechFlowChunk } from "./speechflow-node"

package/speechflow-cli/src/speechflow-node-t2t-google.ts

@@ -5,7 +5,7 @@
 */
 
 /*  standard dependencies  */
-import Stream from "node:stream"
+import Stream                       from "node:stream"
 
 /*  external dependencies  */
 import { TranslationServiceClient } from "@google-cloud/translate"

package/speechflow-cli/src/speechflow-node-t2t-opus.ts

@@ -68,18 +68,19 @@ export default class SpeechFlowNodeT2TOPUS extends SpeechFlowNode {
         }, 1000)
 
         /*  instantiate Transformers engine and model  */
-        const pipeline = Transformers.pipeline("translation", model, {
-            cache_dir: path.join(this.config.cacheDir, "transformers"),
-            dtype:     "q4",
-            device:    "auto",
-            progress_callback: progressCallback
-        })
-        this.translator = await pipeline
-        if (this.translator === null)
-            throw new Error("failed to instantiate translator pipeline")
-
-        /*  clear progress interval again  */
-        clearInterval(interval)
+        try {
+            const pipeline = Transformers.pipeline("translation", model, {
+                cache_dir: path.join(this.config.cacheDir, "transformers"),
+                dtype:     "q4",
+                device:    "auto",
+                progress_callback: progressCallback
+            })
+            this.translator = await pipeline
+        }
+        finally {
+            /*  clear progress interval again  */
+            clearInterval(interval)
+        }
 
         /*  provide text-to-text translation  */
         const translate = async (text: string) => {
@@ -120,17 +121,17 @@ export default class SpeechFlowNodeT2TOPUS extends SpeechFlowNode {
 
     /*  close node  */
     async close () {
-        /*  shutdown stream  */
-        if (this.stream !== null) {
-            await util.destroyStream(this.stream)
-            this.stream = null
-        }
-
         /*  shutdown Transformers  */
         if (this.translator !== null) {
             this.translator.dispose()
             this.translator = null
         }
+
+        /*  shutdown stream  */
+        if (this.stream !== null) {
+            await util.destroyStream(this.stream)
+            this.stream = null
+        }
     }
 }
 

package/speechflow-cli/src/speechflow-node-t2t-punctuation.ts

@@ -74,7 +74,7 @@ export default class SpeechFlowNodeT2TPunctuation extends SpeechFlowNode {
                 "Gib KEINE Erklärungen.\n" +
                 "Gib KEINE Einleitung.\n" +
                 "Gib KEINE Kommentare.\n" +
-                "Gib KEINE Preamble.\n" +
+                "Gib KEINE Präambel.\n" +
                 "Gib KEINEN Prolog.\n" +
                 "Gib KEINEN Epilog.\n" +
                 "Ändere NICHT die Wörter.\n" +

package/speechflow-cli/src/speechflow-node-t2t-spellcheck.ts

@@ -66,7 +66,7 @@ export default class SpeechFlowNodeT2TSpellcheck extends SpeechFlowNode {
                 "Gib KEINE Erklärungen.\n" +
                 "Gib KEINE Einleitung.\n" +
                 "Gib KEINE Kommentare.\n" +
-                "Gib KEINE Preamble.\n" +
+                "Gib KEINE Präambel.\n" +
                 "Gib KEINEN Prolog.\n" +
                 "Gib KEINEN Epilog.\n" +
                 "Ändere NICHT die Grammatik.\n" +

package/speechflow-cli/src/speechflow-node-t2t-subtitle.ts

@@ -191,7 +191,7 @@ export default class SpeechFlowNodeT2TSubtitle extends SpeechFlowNode {
                 for (const block of blocks) {
                     const lines = block.trim().split(/\r?\n/)
                     if (lines.length < 2) {
-                        this.log("warning", "SRT block contains less than 2 lines")
+                        this.log("warning", "SRT block contains fewer than 2 lines")
                         continue
                     }
 
@@ -231,7 +231,7 @@ export default class SpeechFlowNodeT2TSubtitle extends SpeechFlowNode {
                 for (const block of blocks) {
                     const lines = block.trim().split(/\r?\n/)
                     if (lines.length < 1) {
-                        this.log("warning", "VTT block contains less than 1 line")
+                        this.log("warning", "VTT block contains fewer than 1 line")
                         continue
                     }
 
@@ -394,15 +394,18 @@ export default class SpeechFlowNodeT2TSubtitle extends SpeechFlowNode {
                     h.response({}).code(204)
             })
 
+            /*  start HAPI server  */
             await this.hapi.start()
             this.log("info", `HAPI: started REST/WebSocket network service: http://${this.params.addr}:${this.params.port}`)
 
+            /*  helper to emit chunks to WebSocket peers  */
             const emit = (chunk: SpeechFlowChunk) => {
                 const data = JSON.stringify(chunk)
                 for (const info of wsPeers.values())
                     info.ws.send(data)
             }
 
+            /*  establish writable stream  */
             this.stream = new Stream.Writable({
                 objectMode:     true,
                 decodeStrings:  false,

package/speechflow-cli/src/speechflow-node-t2t-summary.ts

@@ -60,7 +60,7 @@ export default class SpeechFlowNodeT2TSummary extends SpeechFlowNode {
                 "Gib KEINE Erklärungen.\n" +
                 "Gib KEINE Einleitung.\n" +
                 "Gib KEINE Kommentare.\n" +
-                "Gib KEINE Prämbel.\n" +
+                "Gib KEINE Präambel.\n" +
                 "Gib KEINEN Prolog.\n" +
                 "Gib KEINEN Epilog.\n" +
                 "Komme auf den Punkt.\n" +

package/speechflow-cli/src/speechflow-node-x2x-filter.ts

@@ -120,6 +120,8 @@ export default class SpeechFlowNodeX2XFilter extends SpeechFlowNode {
                     val1 = chunk.timestampStart.toMillis()
                 else if (self.params.var === "time:end")
                     val1 = chunk.timestampEnd.toMillis()
+                else
+                    val1 = undefined
                 if (comparison(val1, self.params.op, val2)) {
                     self.log("info", `[${self.params.name}]: passing through ${chunk.type} chunk`)
                     this.push(chunk)

package/speechflow-cli/src/speechflow-node-xio-exec.ts

@@ -198,6 +198,7 @@ export default class SpeechFlowNodeXIOExec extends SpeechFlowNode {
             this.subprocess.removeAllListeners("error")
             this.subprocess.removeAllListeners("exit")
 
+            /*  clear subprocess reference  */
             this.subprocess = null
         }
 

package/speechflow-cli/src/speechflow-node-xio-file.ts

@@ -54,7 +54,7 @@ export default class SpeechFlowNodeXIOFile extends SpeechFlowNode {
     /*  open node  */
     async open () {
         /*  determine how many bytes we need per chunk when
-            the chunk should be of the required duration/size */
+            the chunk should be of the required duration/size  */
         const highWaterMarkAudio = (
             this.config.audioSampleRate *
             (this.config.audioBitDepth / 8)
@@ -139,11 +139,10 @@ export default class SpeechFlowNodeXIOFile extends SpeechFlowNode {
                             const payload = Buffer.isBuffer(chunk.payload) ?
                                 chunk.payload : Buffer.from(chunk.payload)
                             const seekPosition = chunk.meta.get("chunk:seek") as number | undefined
-                            if (seekPosition !== undefined) {
+                            if (seekPosition !== undefined)
                                 /*  seek to specified position and write (overload)  */
                                 fs.write(self.fd!, payload, 0, payload.byteLength, seekPosition, callback)
-                            }
-                            else {
+                            else
                                 /*  append at current position  */
                                 fs.write(self.fd!, payload, 0, payload.byteLength, writePosition, (err) => {
                                     if (err)
@@ -153,7 +152,6 @@ export default class SpeechFlowNodeXIOFile extends SpeechFlowNode {
                                         callback()
                                     }
                                 })
-                            }
                         },
                         final (callback) {
                             callback()

package/speechflow-cli/src/speechflow-node-xio-mqtt.ts

@@ -21,8 +21,8 @@ export default class SpeechFlowNodeXIOMQTT extends SpeechFlowNode {
     public static name = "xio-mqtt"
 
     /*  internal state  */
-    private broker: MQTT.MqttClient | null = null
-    private clientId: string = (new UUID(1)).format()
+    private broker:     MQTT.MqttClient | null                   = null
+    private clientId:   string                                   = (new UUID(1)).format()
     private chunkQueue: util.SingleQueue<SpeechFlowChunk> | null = null
 
     /*  construct node  */

package/speechflow-cli/src/speechflow-node-xio-vban.ts

@@ -29,11 +29,11 @@ export default class SpeechFlowNodeXIOVBAN extends SpeechFlowNode {
     public static name = "xio-vban"
 
     /*  internal state  */
-    private server:     VBANServer                        | null = null
-    private chunkQueue: util.SingleQueue<SpeechFlowChunk> | null = null
-    private frameCounter                                         = 0
-    private targetAddress                                        = ""
-    private targetPort                                           = 0
+    private server:        VBANServer                        | null = null
+    private chunkQueue:    util.SingleQueue<SpeechFlowChunk> | null = null
+    private frameCounter                                            = 0
+    private targetAddress                                           = ""
+    private targetPort                                              = 0
 
     /*  construct node  */
     constructor (id: string, cfg: { [ id: string ]: any }, opts: { [ id: string ]: any }, args: any[]) {

package/speechflow-cli/src/speechflow-node-xio-webrtc.ts

@@ -154,6 +154,7 @@ export default class SpeechFlowNodeXIOWebRTC extends SpeechFlowNode {
             this.pcmBuffer = this.pcmBuffer.subarray(this.pcmBuffer.length - maxBufferSize)
         }
 
+        /*  process full Opus frames from buffer  */
         while (this.pcmBuffer.length >= this.OPUS_FRAME_BYTES) {
             const frame = this.pcmBuffer.subarray(0, this.OPUS_FRAME_BYTES)
             this.pcmBuffer = this.pcmBuffer.subarray(this.OPUS_FRAME_BYTES)
@@ -418,6 +419,7 @@ export default class SpeechFlowNodeXIOWebRTC extends SpeechFlowNode {
                 const isPublisher  = hasSendonly || hasSendrecv
                 const isViewer     = hasRecvonly
 
+                /*  handle protocol based on mode  */
                 if (self.params.mode === "r" && isPublisher)
                     /*  in read mode, accept WHIP publishers  */
                     await self.handleWHIP(res, body)