DeepSpeech

Mozilla DeepSpeech is a speech-to-text engine powered by TensorFlow for machine learning and derived from Baidu's Deep Speech research paper. This test profile times the speech-to-text process for a roughly three minute audio recording.

To run this test with the Phoronix Test Suite, the basic command is: phoronix-test-suite benchmark deepspeech.

Project Site

github.com

Test Created

8 December 2019

Test Maintainer

Michael Larabel 

Test Type

Processor

Average Install Time

17 Seconds

Average Run Time

5 Minutes, 54 Seconds

Accolades

5k+ Downloads

Supported Platforms

Supported Architectures

x86_64

Public Result Uploads *Reported Installs **Reported Test Completions **Test Profile Page Views ***OpenBenchmarking.orgEventsDeepSpeech Popularity Statisticspts/deepspeech2019.122020.012020.022020.032020.042020.052020.062020.072020.082020.092020.102020.112020.122021.012021.022021.032021.042021.052021.062021.07400800120016002000
* Uploading of benchmark result data to OpenBenchmarking.org is always optional (opt-in) via the Phoronix Test Suite for users wishing to share their results publicly.
** Data based on those opting to upload their test results to OpenBenchmarking.org and users enabling the opt-in anonymous statistics reporting while running benchmarks from an Internet-connected platform.
*** Test profile page view reporting began March 2021.
Data current as of 29 July 2021.

Revision History

pts/deepspeech-1.0.0   [View Source]   Sun, 08 Dec 2019 12:41:29 GMT
Initial commit of Mozilla DeepSpeech

Suites Using This Test

Machine Learning

Creator Workloads

HPC - High Performance Computing

Speech

Single-Threaded

Telephony


Performance Metrics

Analyze Test Configuration:

DeepSpeech 0.6

Acceleration: CPU

OpenBenchmarking.org metrics for this test profile configuration based on 1,304 public results since 8 December 2019 with the latest data as of 26 July 2021.

Below is an overview of the generalized performance for components where there is sufficient statistically significant data based upon user-uploaded results. It is important to keep in mind particularly in the Linux/open-source space there can be vastly different OS configurations, with this overview intended to offer just general guidance as to the performance expectations.

Component
Percentile Rank
# Compatible Public Results
Seconds (Average)
94th
14
57 +/- 4
90th
3
60 +/- 1
76th
5
67 +/- 5
76th
6
67 +/- 1
76th
12
67 +/- 1
76th
12
67 +/- 3
Mid-Tier
75th
> 68
75th
36
68 +/- 9
72nd
15
69 +/- 6
72nd
7
69 +/- 1
71st
14
70 +/- 3
71st
3
70 +/- 1
70th
3
71 +/- 2
70th
5
71 +/- 3
70th
12
71 +/- 4
70th
10
71 +/- 2
65th
11
73 +/- 6
59th
10
76 +/- 4
58th
20
77 +/- 3
57th
12
78 +/- 4
57th
15
78 +/- 3
54th
28
79 +/- 1
51st
12
80 +/- 10
51st
5
80 +/- 11
Median
50th
80
48th
16
81 +/- 9
48th
4
81 +/- 8
40th
8
85 +/- 2
38th
12
90 +/- 6
36th
3
93 +/- 2
35th
9
95 +/- 4
35th
5
95 +/- 2
33rd
12
99 +/- 7
27th
6
109 +/- 11
26th
7
110 +/- 1
Low-Tier
25th
> 111
23rd
34
115 +/- 7
22nd
5
116 +/- 11
21st
10
118 +/- 7
19th
3
121 +/- 8
17th
8
123 +/- 5
14th
4
129 +/- 4
14th
13
130 +/- 9
13th
15
131 +/- 12
10th
3
138 +/- 4
10th
17
139 +/- 11
2nd
4
271 +/- 25
OpenBenchmarking.orgDistribution Of Public Results - Acceleration: CPU1304 Results Range From 49 To 377 Seconds496785103121139157175193211229247265283301319337355373391100200300400500

Based on OpenBenchmarking.org data, the selected test / test configuration (DeepSpeech 0.6 - Acceleration: CPU) has an average run-time of 5 minutes. By default this test profile is set to run at least 3 times but may increase if the standard deviation exceeds pre-defined defaults or other calculations deem additional runs necessary for greater statistical accuracy of the result.

OpenBenchmarking.orgMinutesTime Required To Complete BenchmarkAcceleration: CPURun-Time510152025Min: 1 / Avg: 4.08 / Max: 19

Based on public OpenBenchmarking.org results, the selected test / test configuration has an average standard deviation of 0.6%.

OpenBenchmarking.orgPercent, Fewer Is BetterAverage Deviation Between RunsAcceleration: CPUDeviation246810Min: 0 / Avg: 0.59 / Max: 6

Does It Scale Well With Increasing Cores?

No, based on the automated analysis of the collected public benchmark data, this test / test settings does not generally scale well with increasing CPU core counts. Data based on publicly available results for this test / test settings, separated by vendor, result divided by the reference CPU clock speed, grouped by matching physical CPU core count, and normalized against the smallest core count tested from each vendor for each CPU having a sufficient number of test samples and statistically significant data.

AMDIntelOpenBenchmarking.orgRelative Core Scaling To BaseDeepSpeech CPU Core ScalingAcceleration: CPU2468121618243248641.01192.02383.03574.04765.0595

Notable Instruction Set Usage

Notable instruction set extensions supported by this test, based on an automatic analysis by the Phoronix Test Suite / OpenBenchmarking.org analytics engine.

Instruction Set
Support
Instructions Detected
SSE2 (SSE2)
Used by default on supported hardware.
 
UNPCKLPD CVTPD2PS CVTSI2SD DIVSD CVTTSD2SI XORPD UCOMISD MOVAPD MULSD ADDSD CMPNLESD ANDPD ANDNPD ORPD SUBSD CMPLTSD CVTPS2PD SQRTSD CMPNLTSD CMPLESD MAXSD MOVDQU
SSSE3 (SSSE3)
Used by default on supported hardware.
 
PSHUFB PALIGNR
SSE 4.2 (SSE4_2)
Used by default on supported hardware.
Found on Intel processors since at least 2010.
Found on AMD processors since Bulldozer (2011).

 
POPCNT
Used by default on supported hardware.
Found on Intel processors since Sandy Bridge (2011).
Found on AMD processors since Bulldozer (2011).

 
VZEROUPPER VEXTRACTF128 VINSERTF128 VBROADCASTF128 VBROADCASTSD VBROADCASTSS VPERM2F128 VPERMILPD VPERMILPS VZEROALL
Used by default on supported hardware.
Found on Intel processors since Haswell (2013).
Found on AMD processors since Excavator (2016).

 
VPERM2I128 VEXTRACTI128 VPERMD VPBROADCASTQ VPERMQ VPBLENDD VPBROADCASTD
Used by default on supported hardware.
Found on Intel processors since Westmere (2011).
Found on AMD processors since Bulldozer (2011).

 
AESENC AESENCLAST AESDEC AESDECLAST AESIMC AESKEYGENASSIST
Vector AES (VAES)
Used by default on supported hardware.
Found on Intel processors since Ice Lake (2019).
Found on AMD processors since Zen 3 (2020).

 
VAESENC VAESENCLAST
Last automated analysis: 10 May 2021

This test profile binary relies on the shared libraries libdeepspeech.so, libdl.so.2, libm.so.6, libgomp.so.1, libpthread.so.0, libc.so.6, librt.so.1.

Recent Test Results

OpenBenchmarking.org Results Compare

4 Systems - 78 Benchmark Results

AMD Ryzen 9 5900HX - ASUS G513QY v1.0 - AMD Renoir Root Complex

Arch Linux - 5.13.4-arch1-1 - GNOME Shell 40.3

3 Systems - 78 Benchmark Results

AMD Ryzen 9 5900HX - ASUS G513QY v1.0 - AMD Renoir Root Complex

Arch Linux - 5.13.4-arch1-1 - GNOME Shell 40.3

2 Systems - 75 Benchmark Results

AMD Ryzen 9 5900HX - ASUS G513QY v1.0 - AMD Renoir Root Complex

Ubuntu 21.04 - 5.13.2 - GNOME Shell 3.38.4

3 Systems - 189 Benchmark Results

AMD Ryzen 9 5900HX - ASUS G513QY v1.0 - AMD Renoir Root Complex

Ubuntu 21.04 - 5.13.0-051300-generic - GNOME Shell 3.38.4

2 Systems - 189 Benchmark Results

AMD Ryzen 9 5900HX - ASUS G513QY v1.0 - AMD Renoir Root Complex

Ubuntu 21.04 - 5.13.0-051300-generic - GNOME Shell 3.38.4

3 Systems - 240 Benchmark Results

AMD EPYC 72F3 8-Core - Supermicro H12SSL-i v1.01 - AMD Starship

Ubuntu 21.04 - 5.11.0-16-generic - GNOME Shell 3.38.4

3 Systems - 240 Benchmark Results

AMD EPYC 72F3 8-Core - Supermicro H12SSL-i v1.01 - AMD Starship

Ubuntu 21.04 - 5.11.0-16-generic - GNOME Shell 3.38.4

2 Systems - 339 Benchmark Results

AMD Ryzen Threadripper 3990X 64-Core - Gigabyte TRX40 AORUS PRO WIFI - AMD Starship

Pop 21.04 - 5.11.0-7620-generic - GNOME Shell 3.38.4

1 System - 88 Benchmark Results

AMD Ryzen 5 3600 6-Core - Gigabyte X570 AORUS PRO - AMD Starship

ManjaroLinux 21.1.0 - 5.13.1-3-MANJARO - X Server 1.20.11

2 Systems - 354 Benchmark Results

AMD Ryzen Threadripper 3990X 64-Core - Gigabyte TRX40 AORUS PRO WIFI - AMD Starship

Pop 20.04 - 5.11.0-7620-generic - GNOME Shell 3.36.7

1 System - 352 Benchmark Results

AMD Ryzen Threadripper 3990X 64-Core - Gigabyte TRX40 AORUS PRO WIFI - AMD Starship

Pop 20.04 - 5.11.0-7620-generic - GNOME Shell 3.36.7

Most Popular Test Results

OpenBenchmarking.org Results Compare

3 Systems - 268 Benchmark Results

Intel Core i5-2520M - HP 161C - Intel 2nd Generation Core DRAM

Ubuntu 18.04 - 4.18.0-20-generic - GNOME Shell 3.28.3

16 Systems - 119 Benchmark Results

AMD EPYC 7R32 48-Core - Supermicro H11SSL-i v2.00 - AMD Starship

Ubuntu 19.10 - 5.3.0-64-generic - GNOME Shell 3.34.1

2 Systems - 535 Benchmark Results

AMD Ryzen 5 4500U - LENOVO LNVNB161216 - AMD Renoir Root Complex

Ubuntu 20.04 - 5.9.0-050900rc7daily20201002-generic - GNOME Shell 3.36.4

11 Systems - 217 Benchmark Results

AMD Ryzen 5 2600 Six-Core - ASUS ROG CROSSHAIR VIII HERO - AMD 17h

Ubuntu 20.04 - 5.9.0-050900-generic - GNOME Shell 3.36.4

12 Systems - 593 Benchmark Results

AMD Ryzen 5 3600XT 6-Core - MSI X470 GAMING M7 AC - AMD Starship

Ubuntu 20.04 - 5.8.0-050800daily20200622-generic - GNOME Shell 3.36.2

4 Systems - 55 Benchmark Results

AMD Ryzen 3 3200U - MOTILE PF4PU1F - AMD Device 15d0

Ubuntu 18.04 - 5.3.0-28-generic - GNOME Shell 3.28.4

2 Systems - 151 Benchmark Results

Intel Xeon E-2278GEL - Logic Supply RXM-181 - Intel Cannon Lake PCH

Ubuntu 20.10 - 5.8.0-41-generic - GNOME Shell 3.38.2

3 Systems - 174 Benchmark Results

Intel Core i9-10900K - Gigabyte Z490 AORUS MASTER - Intel Comet Lake PCH

Fedora 32 - 5.6.6-300.fc32.x86_64 - GNOME Shell 3.36.1

12 Systems - 229 Benchmark Results

AMD Ryzen 7 3800XT 8-Core - ASUS ROG CROSSHAIR VIII HERO - AMD Starship

Ubuntu 20.04 - 5.9.0-050900-generic - GNOME Shell 3.36.4

2 Systems - 1708 Benchmark Results

AMD Ryzen 3 3300X 4-Core - ASRock X570 Pro4 - AMD Starship

Ubuntu 20.04 - 5.7.0-rc6-amd-energy - GNOME Shell 3.36.2

11 Systems - 72 Benchmark Results

Intel Core i9-10980XE - Gigabyte X299X DESIGNARE 10G - Intel Sky Lake-E DMI3 Registers

CentOS Linux 8 - 4.18.0-147.6.el8.x86_64 - GNOME Shell 3.28.3

Find More Test Results