HTML5 AAC Audio Playback Tests -
xHE-AAC Comparison Page

This page compares xHE-AAC to HE-AAC to show the improvements in coding efficiency, speech quality, dynamic range control, and bitrate switching enabled by xHE-AAC.

To enable playback with legacy browsers, and to show loudness and DRC effects, all the content on this page has been encoded with the listed codec and bitrate, decoded to PCM (uncompressed audio) at a target loudness level of -16 LKFS, and then re-encoded as AAC-LC at a high and transparent bitrate. Note that HE-AAC loudness control is not mandatory and so most HE-AAC content today is not loudness normalized.

Audio Quality Improvements

Improved Speech Quality

Since xHE-AAC includes dedicated speech coding tools, speech quality can be greatly improved at lower bitrates compared to legacy codecs.

Dramatic Film - Netflix "Meridian"

• Stereo version
• Property of Netflix, used with permission
• Listening Tip: listen for less roughness on "So these guys..." at intro and "Mickey Cohen..", improved transients and bandwidth on paper shuffling at beginning and rain at end

Male Speech

• EBU Sound Quality Assessment Material, Track 50
  • Male English Speech
  • C414 cardioid at 65 cm, BBC drama studio, RT₆₀ = 0.3 seconds at midband
  • Narration: L+R Channels
• Listening tips:
  • xHE-AAC speech has less roughness, reverberation, and coding noise. This is sometimes perceived as "double speak" or "ghost voices" on HE-AAC.
  • At bit rates below 24 kbit/s there is a clear difference in audio bandwidth. xHE-AAC sounds less dull and has much more clarity. Sibilants and fricatives ("s" sounds ) are much more intelligible.

Female Speech

• EBU Sound Quality Assessment Material, Track 49
  • Female English Speech
  • C414 cardioid at 65 cm, BBC drama studio, RT₆₀ = 0.3 seconds at midband
  • Narration: L+R Channels
• Listening Tip: xHE-AAC speech has less roughness, reverberation, and coding noise. Sometimes perceived as "double speak" or "ghost voices" on HE-AAC. For HE-AAC the speaker sounds as if she has a sore throat. Listen for less coding noise on "the simplest method is to mix the medicine"

Original Wave Files:
The SQAM test items are the property of the European Broadcast Union and may be downloaded from their website at https://tech.ebu.ch/publications/sqamcd

Transient and Audio Bandwidth Improvement

Music: "Rain"

• Rain
  • Artist/Composer: Navajiva
  • Licensed from Klangarchiv
• Listening Tips:
  • Onsets of the picked guitar are much improved for xHE-AAC. For HE-AAC the individual notes seem to have an echo to them. Also there is warbling or reverb of the guitar on the first notes.
  • For HE-AAC the cello seems to have a rough, erratic noisefloor. xHE-AAC also delivers a much more stable stereo image when both instruments play simultaneously.

Music: "Walking":

• Walking Throughout The City
  • Artist/Composer: Marc Rosenberger
  • Licensed from Klangarchiv
• Listening Tips:
  • xHE-AAC offers more audio bandwidth. Listen to the percussion, particularly the steady shaker in the background. Also the occasional high pitched bell is much more clearly discernible with xHE-AAC.
  • In general for HE-AAC the sound and stereo image seem to be washed out and blurred.

Stereo Image Improvement

Music: "Farewell"

• “The Farewell”
  • Artist/Composer: Albrecht “Abi” Linden
  • Licensed from Klangarchiv
• Listening Tips:
  • xHE-AAC has improved sound on the instrumentals parts as well as on the vocal parts of this piece.
  • For HE-AAC the piano passages between the vocal parts are warbling and seem "unstable". The vocals show the same uncomfortable noise artifacts as heard on the male and female speech items above.

Dynamic Range Control Demonstration

xHE-AAC includes MPEG-D DRC advanced loudness and dynamic range control. In this demonstration, it is used to allow hearing the dialogue in a noisy environment, without the loud portions of the content being played at an objectionable volume. HE-AAC offers only legacy loudness and DRC, without the advanced MPEG-D DRC improvements.

To evaluate the benefits of xHE-AAC's DRC processing, you may want to play this masking noise in the background to simulate a noisy environment:

Masking Noise

Download File

Animated Film - Blender "Sintel"

• Stereo version
• Excerpts edited so that loud and soft scenes are adjacent
• CC BY 3.0 © Copyright Blender Foundation | durian.blender.org

Without Dynamic Range Control
Download File

With Dynamic Range Control
Download File

 

Dramatic Film - Netflix "Meridian"

• Stereo version
• Property of Netflix, used with permission
• Excerpts edited so that loud and soft scenes are adjacent

Without Dynamic Range Control
Download File

With Dynamic Range Control
Download File

 

Bitrate Switching Demonstration

xHE-AAC offers seamless switching over a broad bitrate range, typically 12 to 300 kb/s for stereo coding. Seamless in this case means really seamless. There is a small crossfade of 5-10 ms between the two decoded bitstreams exactly at the specified transition point between them.

This switching allows the use of adaptive streaming protocols such as MPEG DASH or HLS without any special conditioning of fragmented bitstreams as is recommended for HE-AAC adaptive delivery.

Dramatic Film - Netflix "Meridian"

• Stereo version
• Property of Netflix, used with permission
• Switching every two seconds between 24, 32, 64, and 128 kb/s

Download File