• Speech Synthesis LSI Selection Guide

  • Basic knowledge of audio data

  • Listening to compressed audio data

Speech Synthesis LSI Selection Guide (Voice Memory Size Calculator)

Calculate the required voice memory size for your system!

We calculate the voice memory size of your product and propose the best LAPIS Semiconductor Speech Synthesis LSI.

1. Enter the basic settings of the Speech Synthesis LSI

Grade: :
CPU I/F :
Play time :

* The playback time is a guideline at a sampling frequency of 8.0kHz with 4bit ADPCM2.

Memory type :

2.  Input memory specifications and calculate memory size.

Synthesis
method
Number of
phrases
Average
length
of sound
Sampling
frequency
Approximate
memory
size

{{ spec_1_size }}
 Mbit

{{ spec_2_size }}
 Mbit

{{ spec_3_size }}
 Mbit

{{ spec_4_size }}
 Mbit

The total memory size for storing audio data is

{{ totalSize }} Mbit

3. We propose the following LAPIS Semiconductor Speech Synthesis LSI that is most suitable for your system.

  Data
sheets
Supply
Voltage
(V)
Frequency
Band
(MHz)
Oscillator Operating
Temperature
(°C)
ROM
capacity
(bit)
Playback
phrase
count
Maximum
playback
time(s)
CPU
I/F
SP amplifier
output
(W)/class
Mixing
number
(interna)
(ch)
Other PKG Halogen
Free
Support *1
For
automotive *2
Sample
and
Buy

{{ lsi.PartNo }}

{{ lsi.Mixing }}

 
 
 

{{ message }}

  • *1 : Halogen-free compatible mark : Halogen-free compatible products are available. Contact your sales representative for details.
  • *2 :  is for automotive use.   can provide AEC-Q100 qualified products.

Type of synthesis method

Sampling frequency \ synthesis method Description Sound quality Compression ratio Play time (sec)
* When the audio data size is 1Mbit
Bit
rate
Method Human voice Musics Sound effect Fs = 8kHz Fs = 16kHz Fs = 32kHz

(Variable bit length)
HQ-ADPCM *1

LAPIS Semiconductor's unique method.

This method enables high compression while maintaining high sound quality by changing the bit rate to a variable bit length. This is the higher-order method of the following 4-bit ADPCM2.
It is suitable for various sound qualities such as human voice, animal voice, nature sound, music and sound effects.
We recommend this method for new development or when you want to change from another method to one with higher sound quality and higher compression.

High-quality sound High-quality sound High-quality sound 1/5 40 20 10

4bit ADPCM2

LAPIS Semiconductor's unique method.

The audio data is digitized at regular intervals, high compression is achieved by taking the difference from the previous data, and the sound quality is improved by improving the followability of the waveform.
Suitable for expressing human voices, animal voices, and natural sounds.
If you are currently using this method and want to change to a higher sound quality and higher compression method, we recommend the HQ-ADPCM method.

High-quality sound Low sound quality Low sound quality 1/4 32 16 8

8bit non-linearPCM

LAPIS Semiconductor's unique method.

By recording 10bits near the center of the waveform, high compression and high sound quality are realized.
Suitable for expressing low volume sounds or human voice.
If you are currently using this method and want to change to a higher sound quality, higher compression method, we recommend the HQ-ADPCM method.

High-quality sound Medium sound quality Medium sound quality 1/2 16 8 4

8bit PCM

This is a general method.

It is suitable for expressions such as music and sound effects, and for sound effects and pulse-shaped waveforms with sharp changes in waveform.

High-quality sound High-quality sound High-quality sound 1/2 16 8 4

16bit PCM

This is a general method.

It is suitable for expressions such as music and sound effects, and for sound effects and pulse-shaped waveforms with sharp changes in waveform.

High-quality sound High-quality sound High-quality sound 1/1 8 4 2
  • *1 : HQ-ADPCM is Ky's high-quality sound compression technology. Ky's is a registered trademark of Kyushu Institute of Technology.Ky's logo

Phrase description

phrase Description

Sampling Frequency (Fs)

When converting an analog signal such as voice to a digital signal, this is a numerical value that indicates how many times (quantization) should be performed for 1 second of data.
The unit is Hz. For example, sampling is performed 8,000 times per second at 8kHz, and 16,000 times per second at 16kHz.
The higher the sampling frequency, the better the sound quality.
(Sampling frequency of 44.1kHz for music CD and 48kHz for digital broadcasting.)

Bit rate

It is a numerical value when the amount of data uttered per second is expressed in bits.
For data of the same format and sampling frequency, the higher the bit rate, the better the sound quality. However, the amount of data is large.

Number of phrases

The number of types of sound to play.
For speech synthesis LSIs, this is the number of addresses that can be set.
For example, when playing "This is LAPIS Semiconductor", one phrase is used for "This is LAPIS Semiconductor", and two phrases for "This is" and "LAPIS Semiconductor".

Sound effect

Synthesis method HQ-ADPCM 4bit ADPCM2 8bit non-linear PCM 8bit PCM 16bit PCM

32kHz

PLAY PLAY PLAY PLAY PLAY

16kHz

PLAY PLAY PLAY PLAY PLAY

8kHz

PLAY PLAY PLAY PLAY PLAY

Melody

Synthesis method HQ-ADPCM 4bit ADPCM2 8bit non-linear PCM 8bit PCM 16bit PCM

32kHz

PLAY PLAY PLAY PLAY PLAY

16kHz

PLAY PLAY PLAY PLAY PLAY

8kHz

PLAY PLAY PLAY PLAY PLAY

Guidance

Synthesis method HQ-ADPCM 4bit ADPCM2 8bit non-linear PCM 8bit PCM 16bit PCM

16kHz

PLAY PLAY PLAY PLAY PLAY

12kHz

PLAY PLAY PLAY PLAY PLAY

8kHz

PLAY PLAY PLAY PLAY PLAY
  • * All copyrights of the sample audition sound source belong to our company and are protected by the Copyright Act. Unauthorized use of our sample listening sound source is prohibited by law.