Skip to content
Snippets Groups Projects
Select Git revision
  • 1.0.1
  • 1.0.x
  • 1.1.1
  • 1.1.2
  • 1.1.x
  • analysis_fix1
  • analysis_fix2
  • burg_optimization
  • cng_fix
  • cng_fix2
  • draft
  • draft_wip
  • exp-highpass
  • exp_24k_analysis1
  • exp_24k_analysis2
  • exp_24k_analysis3
  • exp_24k_analysis4
  • exp_24k_analysis5
  • exp_analysis
  • exp_analysis6
  • v1.3.1
  • v1.3
  • v1.3-rc2
  • v1.3-rc
  • v1.3-beta
  • v1.2.1
  • v1.2
  • v1.2-rc1
  • v1.2-beta
  • v1.1.5
  • v1.1.4
  • v1.2-alpha2
  • v1.2-alpha
  • v1.1.3
  • v1.1.2
  • v1.1.1
  • v1.1.1-rc
  • draft-ietf-codec-oggopus-06
  • draft-ietf-codec-oggopus-05
  • v1.1.1-beta
40 results
You can move around the graph by using the arrow keys.
Created with Raphaël 2.2.016Jul1587654130Jun292721201918171386432131May2423222118171127Apr222018171530Mar25Adds some smoothing to the energy quantizationTune SILK rate for hybrid with FEC enabledMoving FEC decision to the Opus encoder level (from SILK level)Taking loss rate into account in compute_equiv_rate()Retuning the tf_analysis() lambdaFixes equiv_rate for CBRMinor TF cleanup (tf_sum was useless), plus commentsUse SPREAD_AGGRESSIVE on non-transient hybrid framesDon't use too much temporal resolution on hybrid frames at low rateDisable patch_transient_decision() on hybrid since it's causing artifactsCBR: set gains to their previons values where we're busting the budgetSet pulses to zero if we can't meet rate targetNow that we're preventing huge gains, we no longer need to pretend there'sCBR: lock the gain on a subframe when the number of pulses stops going downmerge back in the LTP gain safety limiterCBR: Using the same gain increase for fixed-point as for floatOops, edited the wrong comment in the previous commitRe-tuning compute_equiv_rate()Getting rid of the SILK penalty for hybrid CBR.Improving CBR for hybrid modeMaking the encoder more aggressive at using wider bandwidthscomment about reduced codelength component in LTP quantizer error metricavoid double if statement for finding best quantized LTP coefsincrease NSQ decision delay from 32 to 40reduced noise floor in noise shaping analysis by 2 dBincrease order of noise shaping filtermore conservative scaling of LTP corrs; remove assert that checks for negative residual energy; discard CB entries leading to such negative energiesClean up: alignment of commentsClean up: replace tabs by spacesremoved prefilterslight clean upundo some loop unrollingpitch analysis (in FIX) now scales the input down at the start, instead of at every stageminor clean up of pointer offsetssimplified computation of LTP coefsAttenuate SILK PLC gain only for unvoiced speechWe should now be able to be less conservative when using CBRMakes the encoder more aggressive about meeting the rate targetUsing more fine-grained complexity settings for SILKFixing bandwidth selection hysteresis