Speech interruptions with Kokoro 0.19 on iOS

[jankammerath]

OS: iOS 18.2, 18.3
Device: iPhone 14, iPad 7th Gen, iPhone simulators

The following generated audio outlines the issue, specifically noticable at 03m 04s:
SoundCloud: sherpa-onnx (iOS) sample

This isolated single generated voice output also shows an "f" being cut off at 0:13:
SoundCloud: sherpa-onnx (iOS) isolated sample

The same sample as above, but generated on HuggingFace with ZeroGPU:
SoundCloud: ZeroGPU sample

The implementation is as follows.

/* init the kokoro model for tts */
self.kokoro = sherpaOnnxOfflineTtsKokoroModelConfig(
    model: model.path(),
    voices: voices.path(),
    tokens: tokens.path(),
    dataDir: dataDir
)

/* setup the serpa onnx offline tts wrapper */
let modelConfig = sherpaOnnxOfflineTtsModelConfig(kokoro: self.kokoro!, debug: 0)
var ttsConfig = sherpaOnnxOfflineTtsConfig(model: modelConfig)
self.tts = SherpaOnnxOfflineTtsWrapper(config: &ttsConfig)

/* generate the audio file in the temp folder */
let text = "Insane in the membrane"
let speed: Float = 1.0
let audioFileName = "/tmp/blabla.wav"
let audio = self.tts?.generate(text: text, sid: Int(voiceId), speed: speed)
_ = audio?.save(filename: tempAudioFile.path())

Is there anything wrong with my implementation or is this just the model? I tried Kokoro 0.19 online at Huggingface and couldn't identify a similar issue. Therefore I am not sure if this is just the model, my code or a bug. Hence, reporting it here for potential investigation.