TWELP 1200 bps Vocoder - DSP Innovations

Details: Published: 2016 November 09; Last Updated: 2026 August 10

offers speech quality and intelligibility that were previously achievable only at nearly double the bitrate. Below are the results of comparative tests with the MELPe 2400 bps vocoder, which operates at twice the bitrate.

Deployed in Digital HF Radio, tactical communications, secure voice, satellite communications, and other bandwidth-constrained communication applications.

TWELP Technology

Advanced Speech Coding Approach: Pitch-synchronous processing and proprietary Tri-Wave excitation, providing an alternative to MELPe, AMBE+2 and ACELP-based approaches.
Mature & Proven: Nearly two decades of continuous development and real-world deployment in professional communication systems.
Speech & Beyond: High-quality speech plus improved reproduction of tones, sirens, music and other non-speech signals.

Speech Quality

TWELP 1200 was compared with MELPe at 2400 and 1200 bps using an updated 8 kHz test set derived from the ITU-T P.50 speech database in 20 languages.

Note:
The original ITU-T P.50 speech files were resampled to 8 kHz. Files for each language were combined and inter-speech pauses were shortened to minimize the effect of silence on the evaluation results.

Speech quality was objectively evaluated using ITU-T P.862 (PESQ):

Across 20 languages, TWELP 1200 bps consistently outperforms MELPe at the same 1200 bps bitrate. Exact PESQ scores are shown below.

Language	MELPe 2400 bps	TWELP 1200 bps	MELPe 1200 bps
American	2.967	2.899	2.711
Arabic	3.024	2.868	2.722
British	2.885	2.791	2.689
Chinese	2.871	2.939	2.607
Danish	2.946	2.963	2.720
Dutch	2.856	2.772	2.644
Finnish	2.820	2.791	2.630
French	3.018	2.943	2.818
German	2.986	2.923	2.793
Greek	2.950	2.921	2.698
Hindi	3.070	2.968	2.847
Hungarian	3.060	2.944	2.793
Italian	3.213	3.092	2.983
Japanese	3.133	3.021	2.898
Norwegian	3.007	2.948	2.745
Polish	3.027	2.943	2.758
Portuguese	3.113	3.009	2.894
Russian	2.863	2.849	2.689
Spanish	3.046	2.928	2.816
Swedish	3.119	3.019	2.907
Average	2.999	2.927	2.768
The results confirm that TWELP 1200 bps provides speech quality close to that of MELPe 2400 bps while operating at half the bitrate.

Speech Intelligibility

Speech intelligibility was evaluated using three complementary objective metrics: ASR-based Speech Intelligibility, STOI and ESTOI.

STOI and ESTOI were calculated using the updated 8 kHz test set described above. ASR-based Speech Intelligibility was evaluated separately using the original ITU-T P.50 speech files at their original 16 kHz sampling rate.

ASR reference transcripts were generated by whisper.cpp with the Whisper large-v3 model directly from the original, unprocessed 16 kHz P.50 speech files. For vocoder evaluation, the same speech files were converted to 8 kHz, processed by the vocoder, converted back to 16 kHz and transcribed using the same ASR system.
ASR-based Speech Intelligibility is calculated as 100 - CER (%) by comparing each processed-speech transcript with its corresponding ASR reference transcript.
CER is used instead of WER because it measures character-level recognition errors without treating an entire word as incorrect due to a single-character error.
CER scoring ignores letter case, whitespace and punctuation.

Control measurement: the same 16 kHz → 8 kHz → 16 kHz conversion path, but without any vocoder processing, resulted in an overall ASR-based intelligibility of 99.41%. This shows that narrowband conversion alone produces a small measurable reduction in the ASR-based intelligibility measure, even for clean, unencoded speech.
All vocoder results below are scored against the ASR reference transcripts generated from the original, unprocessed 16 kHz speech, not against transcripts from the converted signal.

Across 20 languages, TWELP 1200 bps achieves ASR-based speech intelligibility close to MELPe 2400 bps, while clearly outperforming MELPe at the same 1200 bps bitrate. Exact values are shown below:

Language	MELPe 2400 bps	TWELP 1200 bps	MELPe 1200 bps
American	99.43	98.92	98.51
Arabic	96.31	97.65	94.80
British	100.0	99.41	99.41
Chinese	90.89	97.04	86.70
Danish	93.94	90.73	87.35
Dutch	99.01	98.68	96.75
Finnish	98.20	95.17	92.71
French	98.97	98.93	98.22
German	99.22	98.91	98.83
Greek	98.34	97.01	94.69
Hindi	96.00	93.10	90.03
Hungarian	98.93	98.50	96.86
Italian	99.06	99.18	98.43
Japanese	96.58	96.72	96.58
Norwegian	99.52	98.61	98.55
Polish	98.56	97.48	96.16
Portuguese	99.06	99.38	98.05
Russian	99.78	99.49	99.13
Spanish	95.07	98.95	94.02
Swedish	98.15	96.42	94.25
Overall	97.97	97.29	95.53
Overall ASR-based speech intelligibility: 97.29% for TWELP 1200 bps, versus 97.97% for MELPe 2400 bps and 95.53% for MELPe 1200 bps.

Across 20 languages, TWELP 1200 bps achieves a slightly higher average STOI score than MELPe 2400 bps, while clearly outperforming MELPe at the same 1200 bps bitrate. Exact values are shown below:

Language	MELPe 2400 bps	TWELP 1200 bps	MELPe 1200 bps
American	87.43	88.54	85.20
Arabic	87.70	87.61	84.76
British	85.10	85.87	80.97
Chinese	86.78	87.85	83.74
Danish	87.30	88.87	84.86
Dutch	86.19	86.84	82.66
Finnish	82.94	84.17	80.12
French	87.35	88.40	84.28
German	87.10	88.23	85.01
Greek	86.96	87.84	84.03
Hindi	86.89	88.18	84.38
Hungarian	87.85	87.90	84.72
Italian	87.57	88.19	85.00
Japanese	88.88	87.96	86.01
Norwegian	87.97	88.89	85.12
Polish	87.66	88.18	83.93
Portuguese	87.68	87.90	85.05
Russian	86.89	86.55	82.55
Spanish	86.57	87.25	83.57
Swedish	85.92	86.48	83.47
Average	86.94	87.59	83.97
Average STOI score: 87.59 for TWELP 1200 bps, versus 86.94 for MELPe 2400 bps and 83.97 for MELPe 1200 bps.

ESTOI provides a complementary objective measure of speech intelligibility, extending the STOI analysis with a different treatment of spectro-temporal information:

Across 20 languages, TWELP 1200 bps achieves an average ESTOI score close to MELPe 2400 bps, while clearly outperforming MELPe at the same 1200 bps bitrate. Exact values are shown below:

Language	MELPe 2400 bps	TWELP 1200 bps	MELPe 1200 bps
American	81.43	81.12	78.01
Arabic	82.61	80.87	78.27
British	79.35	78.18	75.02
Chinese	82.03	82.25	78.08
Danish	81.22	82.08	77.70
Dutch	80.88	79.76	76.97
Finnish	77.26	77.01	73.56
French	82.03	81.71	77.91
German	80.14	80.54	76.58
Greek	82.55	81.68	78.51
Hindi	80.48	79.56	76.53
Hungarian	80.57	80.09	76.09
Italian	81.95	80.81	77.88
Japanese	83.96	82.20	80.21
Norwegian	82.53	82.23	78.86
Polish	82.26	81.39	77.75
Portuguese	81.94	80.90	78.04
Russian	80.86	79.11	76.02
Spanish	81.18	80.76	77.57
Swedish	79.60	78.10	75.91
Average	81.24	80.52	77.27
Average ESTOI score: 80.52 for TWELP 1200 bps, versus 81.24 for MELPe 2400 bps and 77.27 for MELPe 1200 bps.

The P.862, STOI/ESTOI, and CER scoring utilities, including the CER source code and ASR result analysis scripts, are available in the “Downloads” section at the bottom of this page.
The open-source whisper.cpp ASR engine and the corresponding Whisper large-v3 model can be obtained separately from their official repositories.
The supplied tools and instructions allow the results presented above to be independently reproduced and verified.

Speech Samples (WAV files)
Independent experts compared TWELP 1200 bps with MELPe 2400 bps and 1200 bps in preference listening tests. Listening preferences were closely divided, but TWELP was preferred overall in comparisons with both MELPe bitrates, with listeners describing its speech as more natural and less synthetic.

Listen to the source speech and the corresponding MELPe 2400 bps, MELPe 1200 bps and TWELP 1200 bps samples below.

For more accurate comparison, we recommend using good-quality headphones or speakers.

The complete P.50 sample sets for all languages are also available in the Downloads section at the bottom of this page.

Language	Source speech	MELPe 2400 bps	MELPe 1200 bps	TWELP 1200 bps
American
Arabic
British
Chinese
Danish
Dutch
Finnish
French
German
Greek
Hindi
Hungarian
Italian
Japanese
Norwegian
Polish
Portuguese
Russian
Spanish
Swedish

Speech & Beyond
Unlike many low-bitrate vocoders optimized primarily for speech, TWELP also provides improved reproduction of non-speech signals, including alert tones, police, ambulance and fire sirens, music and other audio signals.
Combined with natural speech reproduction, this makes TWELP well suited for digital radio and other communication systems where both speech and non-speech signals must be transmitted over a very low-bitrate channel.
Listen to the comparison below:

Source signal	MELPe 2400 bps	MELPe 1200 bps	TWELP 1200 bps

High Robustness To Acoustic Noise

An additional benefit of the TWELP approach is its strong robustness to acoustic noise. Its pitch-synchronous processing, robust pitch estimation and other features of TWELP technology help preserve speech intelligibility when the input speech is contaminated by background noise.
For this comparison, ESTOI was selected as a single objective intelligibility metric to provide a focused and consistent evaluation of speech under acoustic-noise conditions.

Across 20 languages, TWELP 1200 bps achieves average speech intelligibility in acoustic noise close to MELPe 2400 bps, while clearly outperforming MELPe at the same 1200 bps bitrate. Exact ESTOI values are shown below:

Language	MELPe 2400 bps	TWELP 1200 bps	MELPe 1200 bps
American	70.09	69.98	66.92
Arabic	72.88	71.11	69.34
British	68.02	66.98	64.32
Chinese	73.16	73.61	69.84
Danish	68.87	70.18	65.41
Dutch	68.01	67.25	64.64
Finnish	65.96	65.97	62.45
French	72.39	71.57	68.53
German	68.23	68.20	65.00
Greek	71.61	71.77	68.12
Hindi	71.61	67.23	64.62
Hungarian	71.84	71.27	68.20
Italian	70.28	68.67	66.67
Japanese	75.04	73.08	71.71
Norwegian	73.14	73.25	69.64
Polish	71.40	70.51	67.27
Portuguese	71.25	70.31	67.93
Russian	69.19	68.18	65.57
Spanish	71.55	72.64	68.39
Swedish	65.02	64.40	62.05
Average	70.33	69.81	65.77
Average ESTOI score in acoustic noise: 69.81 for TWELP 1200 bps, versus 70.33 for MELPe 2400 bps and 65.77 for MELPe 1200 bps.

In addition to this inherent robustness, TWELP includes an NCSE (Noise Cancellation Speech Enhancement) preprocessor that can further reduce input noise and improve speech quality.
The samples below compare heavily noisy English speech (SNR = 10 dB) processed by MELPe 2400 bps, MELPe 1200 bps and TWELP 1200 bps, first with noise preprocessing disabled and then enabled (MELPe NPP / TWELP NCSE).

NPP NCSE	Input speech (SNR=10dB)	MELPe 2400 bps	MELPe 1200 bps	TWELP 1200 bps
Disabled
Enabled

The NCSE integrated into the TWELP vocoder is described in more detail on the webpage for our standalone product, 'NCSE-AGC Preprocessor'.

High Robustness To The Channel Errors.

The TWELP technology offers highly efficient speech compression by eliminating redundancy while preserving excellent quality and intelligibility. To enhance robustness against transmission errors, we provide specialized versions called TWELP Robust.
These vocoders are based on an effective Joint Source-Channel Coding approach. Each vocoder is equipped with a custom-designed FEC, tailored to its specific characteristics and operational conditions.
TWELP Robust vocoders provide high speech quality simultaneously in noisy channel as well as in noiseless channel. FEC can operate with "soft decisions" as well as with "hard decisions" from a modem. "Soft decisions" mode provides much better robustness in comparison with the "hard decisions" mode.

For all users of our non-robust vocoder versions, we offer the following recommendations.

The diagram below illustrates the sensitivity of bits at the output of the vocoder to communication channel errors.
Essentially, the diagram shows by what percentage speech quality is reduced when a specific bit is distorted. The first bits in order cause catastrophic distortions, while the latter bits have significantly less impact on quality.

	TWELP 1200 bps Vocoder
1	85.63
2	80.35
3	78.96
4	78.84
5	70.49
6	69.52
7	62.32
8	60.21
9	56.90
10	56.48
11	56.26
12	56.04
13	53.50
14	52.61
15	52.51
16	49.73
17	49.06
18	49.02
19	48.20
20	46.46
21	46.41
22	45.55
23	44.52
24	44.52
25	44.42
26	39.35
27	39.33
28	38.07
29	34.91
30	34.19
31	30.94
32	30.13
33	26.61
34	24.64
35	24.33
36	23.25
37	22.81
38	18.09
39	15.37
40	15.16
41	13.97
42	13.82
43	11.46
44	9.91
45	9.89
46	6.63
47	4.93
48	4.35

We strongly recommend using FEC (Forward Error Correction) with unequal protection of the bits in strong accordance with their sensitivity to errors and utilizing 'Soft Decisions' decoding. This will provide the highest robustness of the vocoder against errors in the channel.

Additional Functionalities. The following additional functionalities are developed by DSPINI and integrated into TWELP vocoders:

Noise Cancellation Speech Enhancement (NCSE)
Automatic Gain Control (AGC),
Voice Activity Detector (VAD),
Discontinuous Transmission (DTX),
Tone Detection/Generation (Single tones and Dual tones). The tones are transmitted by the vocoder facilities.

Note:
The Tone Detector/Generator functionalities are not integrated into the code by default but can be added free of charge upon request.

Each functionality has unique features, performance and characteristics, providing significant superiority over any well-known implementations on the market.

Technical Characteristics And Resource Requirements:

Technical characteristics
Bit Rate (bps)	Algorithm	Frame size (ms)	Algorithmic delay (including frame size) (ms)	Sampling rate (kHz)	Signal format	Bit stream format
1200	TWELP	40	60	8	Linear 16-bit PCM	48

Additional functionalities
Name	Functionality	Technical characteristics
Name	Functionality	Name	Value
AGC	Automatic Gain Control	Control range:	0 ... +42 dB
NCSE	Noise Canceller - Speech Enhancer	SNR increasing	20 dB
NCSE	Noise Canceller - Speech Enhancer	Speech quality improvement	> 0.1 PESQ
Tone Detector	Single/Dual tones detection	In accordance with international standards
Tone Generator	Single/Dual tones generation	Special generator, kept continuity of signal (phase and amplitude of signal of previous frame)
DTX	Discontinuous Transmission	Reduces bit rate down to 110 bps in pauses between active speech regions
VAD	Voice Activity Detection	High reliability even with pink noise at an SNR < 0 dB.
CNG	Comfort Noise Generation	Type of noise	"white"
CNG	Comfort Noise Generation	Level	- 60 dB

The NCSE and AGC integrated into the TWELP vocoder are described in more detail on the webpage for our standalone product, 'NCSE-AGC Preprocessor'.

Resources for ARM Cortex-M4 platform
Module	MIPS* peak	Memory (KBytes)
		Program	Data
		Program	Constants	Channel	Heap	Stack
Voice Encoder	96.9	35	169	4.5	4.8	1.0
NCSE	6.4
AGC	0.2
Voice Decoder	14.0
Voice Encoder + Voice Decoder	110.9
Total	117.5

Resources for TI's C64 DSP platform
Module	MIPS* peak	Memory (KBytes)
		Program	Data
		Program	Constants	Channel	Heap	Stack
Voice Encoder	34.6	86	169	4.5	4.8	1.0
NCSE	2.8
AGC	0.1
Voice Decoder	4.0
Voice Encoder + Voice Decoder	38.6
Total	41.5

Resources (estimated) for TI's C55 DSP platform
Module	MIPS* peak	Memory (KBytes)
		Program	Data
		Program	Constants	Channel	Heap	Stack
Voice Encoder	59.0	21	169	4.5	4.8	1.0
NCSE	6.7
AGC	0.2
Voice Decoder	10.0
Voice Encoder + Voice Decoder	69
Total	75.9

* DSPINI continues optimization of the TWELP algorithm and code in order to minimize computational complexity of the vocoder.

Software Integrity and Security. DSPINI guarantees the ABSOLUTE integrity of its software, free from any undocumented features, undeclared capabilities, or hidden functions. Our customers can be assured that none of our software/code contains any secret features or functionalities concealed from the user. If necessary, we are ready to provide the source code of our software products for appropriate certification.
Moreover, our software is available in source code form—you simply need to purchase the appropriate license to use it.

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.

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

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

Customization. The vocoder can be customized under any 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 TWELP technology.

Related Software. This 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:

Datasheet (pdf)
ITU-T P.50 source speech samples (zip)
MELPe 2400 bps speech samples (zip)
MELPe 1200 bps speech samples (zip)
TWELP 1200 bps speech samples (zip)
P.862 and STOI/ESTOI utilities
PC-evaluation package (zip) — on request
User's Guide document (pdf) — on request