All pull requests

Pull requests

• fix: accept config objects in LlmAgent and keep the transcriber in add_task

  #1173·Pranav-Bobde opened Oct 28, 2026·

  ·

• fix: count mark acks that arrive before the playout estimate

  #1172·ANDROIDHASSAN opened Oct 27, 2026·

  ·

• fix: forward Sarvam first_turn_min_speech_frames to the streaming socket (BOLNA-3263)

  #1169·saranshgoyal17 opened Oct 26, 2026Contributor·

  ·

• feat: add hinglish mode for hindi asr BOLNA-3175

  #1168·DeepBhupatkar opened Oct 26, 2026Member·

  ·

• fix: send bool, null and list tool params as json in get and form requests

  #1165·ANDROIDHASSAN opened Oct 25, 2026·

  ·

• fix: escape quotes and newlines in string tool arguments

  #1164·ANDROIDHASSAN opened Oct 25, 2026·

  ·

• feat: add Navana TTS synthesizer

  #1155·arpit-navana opened Oct 22, 2026·

  ·
  22

• fix: ElevenLabs synthesizer and transcriber take a per-call host (BOLNA-3258)

  #1151·san0808 opened Oct 20, 2026Contributor·

  ·
  4

• fix: keep tool arguments separate from execution settings

  #1150·shashank03-dev opened Oct 20, 2026·

  ·
  1

• fix: drop gemini tool calls that name no configured tool

  #1148·ArnavBallinCode opened Oct 19, 2026·

  ·
  1

• Feat jev language switch judge BOLNA-3162

  #1146·DeepBhupatkar opened Oct 19, 2026Member·

  ·
  6

• fix: update context handling in prompt rendering

  #1144·razi7294 opened Oct 19, 2026Contributor·

  ·
  2

• fix(synthesizer): keep Rime turn context stable so reply audio plays

  #1139·fjjjuv opened Oct 15, 2026·

  ·
  1

• fix: send the call language to elevenlabs tts

  #1131·DeepBhupatkar opened Oct 14, 2026Member·

  ·

• fix(synthesizer): keep reading the active synth after a TTS socket reconnect

  #1130·ShaanSolanki opened Oct 14, 2026·

  ·
  1

• fix(synthesizer): end a TTS turn whose socket closes before its end-of-stream

  #1127·pradneshfernandez opened Oct 13, 2026·

  ·
  1

• fix: end the call with the provider's reason when a TTS provider rejects it

  #1124·san0808 opened Oct 13, 2026Contributor·

  ·

• fix: end the call with the provider's reason when an ASR provider rejects the stream

  #1123·san0808 opened Oct 13, 2026Contributor·

  ·

• fix: credit unacked audio that played when the caller barges in

  #1115·san0808 opened Oct 10, 2026Contributor·

  ·

• feat: Jev (TypeSafe) veto on eager-end-of-turn speculative replies

  #1107·tsekino62 opened Oct 7, 2026·

  ·

• feat: forward caller audio to the Realtime transcriber in 40 ms chunks

  #1103·san0808 opened Oct 7, 2026Contributor·

  ·

• fix(models): move the gpt-3.5-turbo defaults to gpt-4.1-mini (OpenAI shuts it down October 23)

  #1102·23821 opened Oct 6, 2026Contributor·

  ·

• Accumulate Sarvam streaming audio_duration across segments

  #1099·web1havv opened Oct 6, 2026Contributor·1 task·

  ·
  1

• fix: key static-node clips to the voice that speaks them on multilingual pools

  #1098·san0808 opened Oct 5, 2026Contributor·

  ·

• fix(synthesizer): keep a multilingual agent speaking after its TTS socket closes

  #1093·pradneshfernandez opened Oct 2, 2026·

  ·
  1