Some our customers request comparison of our vocoders with open-source CODEC2 vocoder.

Here, we present comparison of the CODEC2 (ver 0.8) vocoder with TWELP vocoder for 2400 bps bit rate.
We added into some tests MELPe 2400 bps vocoder- just for additional comparison with this well-known standard vocoder.

Technology Features. The TWELP vocoders are based on newest technology of speech coding called "Tri-Wave Excited Linear Prediction" (TWELP) that was developed by experts of DSPINI. 

TWELP technology is a new class of vocoders that differs from any other LPC-based vocoders by:

  • advance reliable method of pitch estimation
  • pitch-synchronous analysis
  • advance tri-wave model of excitation
  • newest quantization schemes
  • pitch-synchronous synthesis

Thanks to these unique features, TWELP technology provides much better speech quality in comparison with any well-known technologies, including AMBE+2, MELPe, ACELP, etc. on the same bit rate in range from 300 bps up to 9600 bps and more. Moreover, in contrast to other LBR vocoders (like MELPe, etc.) TWELP provides much better quality for non-speech signals like sirens, background music, etc.

CODEC2 is based on old very simple technology SHC (Sinusoidal Harmonic Coding). This technology was popular more 30 years ago and cannot compete with modern technologies because cannot provide high speech quality that would be close to the modern vocoders.

Speech Quality. Here is a comparison of the speech quality in noiseless channel. TWELP 2400 bps vocoder and CODEC2 2400 bps vocoder were tested, using ITU-T P.50 speech base for 20 different languages. ITU-T P.862 utility was used for estimation of the speech quality in PESQ terms:

 
A diagram demonstrates significant superiority of the TWELP as well as MELPe vocoders over CODEC2 vocoder in speech quality. Exact numbers of the PESQ-estimations are presented in the table below.
LanguageTWELP 2400MELPe 2400CODEC2 2400
American 3.247 3.077 2.777
Arabic 3.157 3.053 2.684
British 3.106 3.019 2.596
Chinese 3.192 2.970 2.748
Danish 3.184 3.022 2.735
Dutch 3.004 2.830 2.463
Finnich 2.970 2.791 2.397
French 3.135 3.106 2.767
German 3.121 2.998 2.641
Greek 3.140 3.004 2.625
Hindi 3.152 3.089 2.670
Hungarian 3.192 3.086 2.921
Italian 3.325 3.226 2.889
Japanese 3.261 3.188 2.846
Norwegian 3.173 3.032 2.817
Polish 3.178 3.029 2.594
Portuguese 3.191 3.146 2.832
Russian 3.076 2.952 2.674
Spanish 3.201 3.048 2.958
Swedish 3.203 3.147 2.713
Average3.1603.0412.717

Superiority of the TWELP 2400 and MELPe 2400 over CODEC2 2400 is on average 0.443 and 0.324 PESQ apropriately

Speech Samples (WAV-files). A few independent experts listened CODEC2 2400 bps vocoder in comparison with TWELP 2400 bps vocoder, using method of preferences. All experts preferred TWELP to CODEC2, having noted much better intellegibility and much more natural human-sounding of voice in the TWELP vocoder.

You can play and listen short samples of the source speech as well as the speech processed by these vocoders for any of 20 languages, using links in the table below.
Also, you can download full set of the P.50 samples as zip-files for all languages simultaneously, using the links in the "Downloads" para in a bottom of the page.

LanguageSource speechMELPe 2400 bpsCODEC2 2400 bpsTWELP 2400 bps
American
Arabic
British
Chinese
Danish
Dutch
Finnich
French
German
Greek
Hindi
Hungarian
Italian
Japanese
Norwegian
Polish
Portuguese
Russian
Spanish
Swedish

Superiority In Quality Of The Non-speech Signals. In contrast to other LBR vocoders (MELPe, AMBE+2, etc.), TWELP vocoders provide high quality of non-speech signals, including police, ambulance, fire sirens, etc. This feature in conjunction with high quality natural human-sounding of voice makes TWELP vocoders well suitable for replacement of analog radio by digital radio and also for other applications where high quality transmitting of non-speech signals is relevant along with high quality transmitting of speech signals.

Source typeSource signalMELPe 2400 bpsCODEC2 2400 bpsTWELP 2400 bps
Siren only
With voice

High Robustness To Acoustic Noise. In contrast to other LBR vocoders, TWELP vocoders are well robust to acoustic noise thanks to robust reliable method of pitch estimation and other features of TWELP technology.

Moreover, vocoder includes in-built Noise Cancellation—Speech Enhancement (NCSE) functionality that improves speech quality in noisy acoustic environment. Here, we used our NCSE-preprocessor for CODEC2 too.

NCSE ModeSource signalMELPe 2400 bpsCODEC2 2400 bpsTWELP 2400 bps
Disabled
Enabled

Guarantee And Support.  DSPINI guarantees a quality and accordance of all technical characteristics of the product to requirement of current specifications. Testing and other method of quality control are used for guarantee support.

Who do guarantee a support for "open-source" codec?

Any Platforms.  DSPINI can port any vocoder software into any other DSP, RISC or general- purposes platform inshort time: 1-2 months.

Licensing Terms.  To use TWELP vocoder, customer should obtain a license from DSPINI only.

Customization.  DSPINI can customize any vocoder under specific requirements- other bit rate, frame size, any other robustness to channel errors, etc. Please contact with us for details.

Prospects.  DSPINI is impoving and developing continuously a set of new vocoders with range from 300 bps up to 9600 bps, based on SPR and TWELP technologies.

Related Software.  Any vocoder may be effectively used in a bundle with other DSPINI's products:

  • Linear and acoustic echo cancellers,
  • Multichannel noise cancellers (including two-microphone adaptive array),
  • Wired or radiomodems for any types of channels and bitrates,
  • Other products.

Downloads:

Conclusion.   The open-source CODEC2 vocoder is based on old SHC technology and provides the worst speech quality. However, it's free of charge and may be used in some uncommercial applications or maybe even in some commercial low cost devices.
Of course, we wouldn't recommend to use this primitive vocoder in applications and devices, where quality and reliability are important.