0 Fork 0

mirror of https://github.com/k2-fsa/sherpa-onnx.git synced 2026-11-10 21:11:03 +08:00

Files

T

History

Fangjun Kuang f2e01fe8e3 Widen VAD sample positions to int64_t to fix long-stream SIGSEGV. (#4079 )

Fixes #3635

It is for v2.0.0 and breaks the ABI.

CircularBuffer::head_ and tail_ are absolute sample counters that grow
forever. As int32_t they overflow after 2^31 samples (~37 h at 16 kHz);
tail_ % capacity then goes negative and std::copy() writes to an invalid
iterator, crashing long-running VAD streams.

Widen every absolute stream position to int64_t. Counts and capacities
(Push/Pop/Get's n, Size(), capacity) stay int32_t — they are bounded by
buffer memory, not stream duration.

2026-10-22 23:39:59 +08:00

.gitignore

Support non-streaming zipformer CTC ASR models (#2340 )

2025-07-04 15:57:07 +08:00

README.md

Add Pascal API for Dolphin CTC models (#2096 )

2025-04-03 16:00:22 +08:00

run-vad-with-dolphin-ctc.sh

Add Pascal API for Dolphin CTC models (#2096 )

2025-04-03 16:00:22 +08:00

run-vad-with-moonshine.sh

Add Pascal API for Moonshine models (#1482 )

2024-10-27 12:21:16 +08:00

run-vad-with-sense-voice.sh

Add Lazarus example for generating subtitles using Silero VAD with non-streaming ASR (#1251 )

2024-08-15 22:19:45 +08:00

run-vad-with-whisper.sh

Add Lazarus example for generating subtitles using Silero VAD with non-streaming ASR (#1251 )

2024-08-15 22:19:45 +08:00

run-vad-with-zipformer-ctc.sh

Support non-streaming zipformer CTC ASR models (#2340 )

2025-07-04 15:57:07 +08:00

vad_with_dolphin.pas

Widen VAD sample positions to int64_t to fix long-stream SIGSEGV. (#4079 )

2026-10-22 23:39:59 +08:00

vad_with_moonshine.pas

Widen VAD sample positions to int64_t to fix long-stream SIGSEGV. (#4079 )

2026-10-22 23:39:59 +08:00

vad_with_sense_voice.pas

Widen VAD sample positions to int64_t to fix long-stream SIGSEGV. (#4079 )

2026-10-22 23:39:59 +08:00

vad_with_whisper.pas

Widen VAD sample positions to int64_t to fix long-stream SIGSEGV. (#4079 )

2026-10-22 23:39:59 +08:00

vad_with_zipformer_ctc.pas

Widen VAD sample positions to int64_t to fix long-stream SIGSEGV. (#4079 )

2026-10-22 23:39:59 +08:00

README.md

Introduction

This directory contains examples for how to use the VAD (voice activity detection) with non-streaming speech recognition models.

Directory	Description
run-vad-with-dolphin-ctc.sh	It shows how to use the VAD + Dolphin for speech recognition.
run-vad-with-whisper.sh	It shows how to use the VAD + Whisper for speech recognition.
run-vad-with-sense-voice.sh	It shows how to use the VAD + SenseVoice for speech recognition.
run-vad-with-moonshine.sh	It shows how to use the VAD + Moonshine for speech recognition.

Please refer to non-streaming-asr for more kinds of non-streaming models.