Align max received transmission length buffers to 30 seconds of audio - #715
Merged
Conversation
udpVoiceRxChan was sized 64*0xFFFFF (~67M slots, roughly 1.6GB of slice headers before any packet data) and voiceBytesRxChan 0xFFFFF. Neither bound reflects real traffic - SRS sends 25 packets/sec per transmitter - and an unbounded backlog just hides a stalled consumer instead of surfacing it. Also cap received transmissions at MaxTransmissionDuration (30s): audio beyond that is truncated at the source instead of buffered indefinitely, and the recognizer's sample clamp now derives from the same constant instead of an unrelated ~16.4s magic number. Co-Authored-By: Claude <noreply@anthropic.com>
dharmab
force-pushed
the
reduce-inbound-voice-channel-size
branch
from
July 26, 2026 14:45
c66717e to
6162df1
Compare
dharmab
changed the title
dharmab
enabled auto-merge (rebase)
July 26, 2026 14:47
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Context
Split off from #712, which found this while working in the same area.
udpVoiceRxChanwas sized64*0xFFFFF(~67M slots - roughly 1.6GB of slice headers before any packet data) andvoiceBytesRxChanwas sized0xFFFFF. Neither bound reflects real traffic: SRS sends about 25 packets/sec per transmitter. An unbounded backlog like this just hides a stalled consumer instead of surfacing it.Fix
Replace the magic numbers with named constants sized for many seconds of slack on a busy frequency:
voiceRxBufferSize = 4096for the raw packet channel,transmissionRxBufferSize = 128for the decoded-transmission channel.Testing
🤖 Generated with Claude Code