| Test Play of Speech Samples | Basic Knowledge of Speech Data |
Relationship between Sampling Frequency and Bit Rate
Sampling Frequency
Denotes the number of times per second an analog signal is converted (quantized) into digital. Generally, CDs are sampled at a rate of 44.1kHz. The higher the sampling rate the better the quality.
Ex)
8kHz is sampled in 8,000times/sec.
16kHz is sample in 16,000times/sec.
Bit Rate
Indicates the number of bits transferred per second. The higher the bit rate the better the quality (at a given sampling rate). However, the file size increases as well.
Ex)
The number of bits of 4bit ADPCM2 is 4bit.
The number of bits of 8bit PCM is 8bit.
As shown in the formula below, the playback time can be calculated by dividing the memory capacity by the bitrate.
Time (sec) = Memory Capacity (bits) / Bitrate (bps)
Memory Capacity
This refers to the memory capacity of the speech synthesis LSI. The larger the memory capacity the longer the playback time. The capacity is calculated in bits.
1Mbit=1024Kbits, 1Kbit=1024bits
Example: 4Mbits=4×1024×1024=4194304bits
Features of Synthesis Methods
| Synthesis Methods | Features | Applications |
|---|---|---|
| ADPCM2 | LAPIS Semiconductor's unique compression method. It expresses the sound data in numbers with constant intervals. The data quantity is reduced by the difference from previous data. The compression rate of this method is the highest. |
Ideal for expressing human voice, animal voice, and nature sound. |
| PCM | The highest sound quality but lowest compression rate among 4 methods. | Ideal for expressing music and sound effects. |
| Non-linear 8bit PCM |
LAPIS Semiconductor's unique method. Records in 10bit near the center of waveform. |
Ideal for expressing low human voice. |
Please inquire from
Communication LSI
HOME